Skip to content
Navegación

Change an app

A JSON Merge Patch. A change of what a version is made of (image, start_command, port, ready, env, secrets, size) deploys a new version of an app that has an image; a change of the rest (name, regions, scale, keys, egress, metadata) holds at once and makes none: the nodes that serve the app are told of its scale and its labels. null takes a field back to what it is when nothing is said; image and name cannot be taken away. A patch that changes nothing writes nothing. Takes write; giving the app a secret it does not have takes admin besides. Of the project's default sandbox app (sbx-sandbox, managed_by of the Sandboxes product) the patch takes image alone, a reference or null for the base template: a new version made of it (answered 200 with the app, current_version the new one; the same image again is none), of which the project's creates that name no image are made from then on; anything else is 409 managed_by_sandboxes, as every change of an image's app is.

PATCH
https://api.glac.ai/v1/apps/{app}

Autorización

  • apiKeyHTTPrequired

Parámetros de ruta

  • appstringrequired

    The app's id (app-...) or its name.

Cuerpo de la solicitudJSONopcional

  • nameoptionalstring | null

    1 to 40 lower-case letters, digits and hyphens, beginning with a letter, with no hyphen at its end or two in a row, not beginning with app- or sbx-, and not localhost, internal or shared. It is the name the project's sandboxes on its network find the app by (private), and no other app's network.aliases (409 name_taken). Not given: one is made up.

  • imageoptionalstring | null

    An image reference: ghcr.io/acme/api:1.4, python:3.12, ...@sha256:.... Pulled from a registry on the internet: one that gives the image to anybody, or one that wants a login, with image_pull_secret.

  • image_pull_secretoptionalstring | null

    The name of one of the project's secrets (/v1/secrets) that holds the credentials of the image's registry, for an image that is not public: username:password in printable ASCII, the user name ending at the first colon; a registry that takes a token alone documents the user name that goes with it (AWS for ECR, oauth2accesstoken for Google's, any name for GHCR). The value goes to each node that builds a version, with the image, and there to the tool that pulls it and to nothing else: nothing of it is in an instance, in a template, in an event or in a log, and what a registry says when it refuses it is the version's error, without it (registry_auth_failed). Naming it takes write, and no admin. The secret has to exist (404 secret_not_found) and to hold such a pair (400 invalid_field) when a version is made; a deploy reads it when it asks a node. In a PATCH, null takes it away. Not given: none, and the image is asked of its registry without a login.

    maxLength: 128
  • start_commandoptional

    Not given: the image's own (ENTRYPOINT and CMD).

    Una de estas opciones:

    • StartCommand
    • null
  • start_argsoptional

    The arguments that follow the image's ENTRYPOINT in the place of its CMD, as a Compose service's command does: the node puts the image's ENTRYPOINT in front of them (an image with none runs them alone). Not with start_command, which is the whole start command (400 invalid_field). Not given: the image's CMD.

    Una de estas opciones:

    • array<string>
    • null
  • portoptionalinteger | null

    The port the service listens on. Not given: the one port the image declares (EXPOSE), or 8080 when it declares none or several.

    maximum: 65535 · minimum: 1
  • readyoptional

    Not given: {"tcp": true}.

    Una de estas opciones:

    • Ready
      • httpoptionalstring

        A path, beginning with /: a GET of it on the app's port is answered with a status from 200 to 399.

      • tcpoptionaltrue

        The app's port takes a connection.

      • commandoptional

        A command that succeeds in the instance. It is run by the instance's shell: a list of arguments is written as one line for it.

        Una de estas opciones:

        • string
        • array<string>
    • null
  • envoptionalobject | null

    The environment of the start command, after the image's own. A value may refer to a variable of another service of the project, or of the app itself, or to one of the project's shared variables, alone or in other text: ${{ db.POSTGRES_USER }}, ${{ shared.REGION }}. A service's variables are its env (whose own references are followed), its secrets, and the four the platform gives it (SBX_SERVICE, its name; SBX_PRIVATE_DOMAIN, <name>.internal; SBX_PRIVATE_ADDRESS, its address on the project's network; SBX_PROJECT, the project's id). A version holds what they come to when it is made: the app keeps what was written. A reference to a secret makes the variable that secret, as secrets does: it is the whole of a value, and it takes admin where the app does not have the secret. Every ${{ begins a reference (400 invalid_reference, param at the variable); one that goes round, or refers to a service or a variable that is not there, is refused when a version is made (400 invalid_reference), and one to a secret that is gone too (404 secret_not_found).

  • secretsoptionalobject | null

    Environment variable to the name of one of the project's secrets (/v1/secrets): the secret's value is in the environment of the start command. No variable is in env as well.

  • sizeoptional

    Not given: the free plan's largest sandbox, or the account's plan's if that is smaller.

    Una de estas opciones:

    • Size
      • vcpusoptionalinteger
        minimum: 1
      • memory_bytesoptionalinteger

        An even number of MiB (a whole number of the 2 MiB pages a node loads a machine's memory in), in bytes. Any other size is 400 invalid_field (param size.memory_bytes): no node would make an instance of it, as none makes a sandbox of it (the node's SandboxLimitsCreate.memory_bytes).

        minimum: 134217728
    • null
  • regionsoptionalarray | null

    Not given: all of them.

  • scaleoptional

    Una de estas opciones:

    • Scale
      • min_instancesoptionalinteger

        How many instances of the app are kept awake in a region, whether a request comes or not: the floor. The nodes of the region that serve the app share it as they share max_instances (two nodes and a floor of 2: one on each; a floor of 1: one, on the node that is tried first; a floor of 5: three and two), and a node that is away is made up for by the others. A node makes its part when it is told, not at the first request, and neither stops nor deletes those instances for being idle; they are metered as instances that are awake. 0, which is what it is when nothing is said: none. Lowered, the instances beyond it are idle from that moment, and are stopped and deleted as idle instances are. At most max_instances, where that is given, and at most what the account's plan lets an app ask for in a region (caps.app_instances.limit: 5, 50 and 500): more is refused, 403 plan_cap_reached.

        minimum: 0
      • max_instancesoptionalinteger | null

        How many instances the app has at the most in a region: the nodes of the region that serve the app share it. At most what the account's plan lets an app ask for (caps.app_instances.limit: 5 on Free, 500 on Paid): more is refused, 403 plan_cap_reached, and what an app asked for on a plan that let it is kept, and held to the plan's number. null: the plan's number for an app that names none (caps.app_instances.default: 2 on Free, 50 on Paid), or min_instances where that is more. What the nodes are told is no more than the account has room for awake beside everything else of it that is awake, nor less than the app has awake.

        minimum: 1
      • concurrencyoptionalinteger | null

        A cap on what one instance has in hand at a time: the requests the edges sent it, and the connections it has open on its project's network (through its service's address, which the node shares out among the instances connection by connection, or to its own). An instance with that many is full, and another is made for more. null, which is what it is when nothing is said: no cap. A node then adds an instance when the ones it has are near their size in CPU or memory, and an app does not have to guess a number; a service that takes one request at a time gives one.

        maximum: 1000 · minimum: 1
      • sleep_after_msoptionalinteger | null

        The most an instance waits without a request before it stops, in milliseconds. Below it, its node picks the wait by the app's own traffic: the wait that would have cost least over the app's last gaps between a request's end and the next one, a second up and idle against a stop and the resume after it in CPU time, so that an app whose requests come closer together than a stop pays for stays up through them, and one whose requests are far apart stops as soon as its last answer has ended (0: stops at once, always). null, which is what it is when nothing is said: the platform's most, three seconds. An app that said 100, the default before 2026-10-28, says null from then on. One that stays stopped is deleted by its node after delete_after_ms, and the next request that needs an instance gets a new one; the instances min_instances keeps awake are neither stopped nor deleted.

        maximum: 600000 · minimum: 0
      • sleep_notice_msoptionalinteger | null

        How long before an instance stops for being idle the programs connected to its guest's /run/sbx/awake are sent the line sleep N, in milliseconds: they flush what they hold, close their connections, or raise a hold, which puts the stop off. The stop comes that long after, whether they answer or not; where nobody is connected, it is not put off. A keyed instance is told as well. 0 or null: not told (the default).

        maximum: 60000 · minimum: 0
      • sleep_hold_max_msoptionalinteger

        How long the holds a program in an instance raises keep it up after its time idle has run out, in milliseconds: a program connects to /run/sbx/awake and sends + before it answers a request whose work goes on after the answer, and - (or closes the connection, or dies) when that work is done; the instance stops sleep_after_ms after the last hold is lowered. Past this time it stops whatever holds it, the programs connected told sleep N cap first where sleep_notice_ms asks for a notice. Held, it runs and is metered as running. 0: holds are not heeded. Null: 30000 (the default).

        maximum: 900000 · minimum: 0
      • suspend_after_msoptionalinteger | null

        How long a stopped instance stays paused before its memory is saved and it sleeps for real, in milliseconds: paused, the next request is answered in a millisecond or two and its memory is metered; asleep, it wakes in a few more and only what it keeps is. Its node holds it to the longest it holds any instance paused (ten minutes); the host's memory running short, a slot another machine needs and a sleep somebody asks for put it to sleep sooner. Null (the default): its node's longest. A database's suspend_after_seconds is its app's.

        maximum: 86400000 · minimum: 1000
      • wake_daily_atoptionalstring | null

        A time of day in UTC, HH:MM, at which the app's keyed instances and its stateful service's instance that rest are woken every day, for work of their own that waits for a wake (a database's maintenance, say); the instances the nodes make and delete by the app's requests are woken by requests alone. The wake is no activity: what runs then is, and once that is done the instance rests again. A stateful service's instance is woken besides for the next run of its PostgreSQL server's pg_cron jobs, whatever this says. Null (the default): none. A database's wake_daily_at is its app's.

        pattern: ^([01][0-9]|2[0-3]):[0-5][0-9]$
      • delete_after_msoptionalinteger | null

        After how long idle an instance is deleted, counted from where its stop is counted from: the next request that needs one gets a new instance, a clone of the version's template, without what the deleted one had gathered in its memory and on its disk (keep what is to last in a volume or a database). A second to an hour, the platform's time, which is the most: more is refused (400 invalid_field), and an instance that is to stay warm is one of min_instances. null, which is what it is when nothing is said: the hour of the nodes (--app-delete-after-seconds). A change holds for the instances the app has as for the ones made after: each node judges every instance by the app's time as it is, once a second at the least. Not for the app's keyed instances, which keys.delete_after_idle_ms is for.

        maximum: 3600000 · minimum: 1000
    • null
  • keysoptional

    Una de estas opciones:

    • AppKeys
      • fromoptional

        Where a request for the app carries its key. A request that carries a key goes to the key's instance (made for it where the key names none: POST /v1/apps/{app}/instances), and one that carries none goes as any request for the app does. A change holds at once, for the requests that come after it. null: requests carry none, and keys are named through /v1/apps/{app}/instances alone.

        Una de estas opciones:

        • AppKeysFrom
        • null
      • upgrade_on_deployoptionalboolean | null

        Every deploy of the app makes a batch of upgrades of its keys' instances to the new version, each once it sleeps (POST /v1/apps/{app}/upgrades with its defaults; slice U4), which stops the batch an earlier deploy made, so the keys follow the newest. false (null is the same), which is what it is when nothing is said: a key's instance stays on its version until it is upgraded.

      • delete_after_idle_msoptionalinteger | null

        After how long asleep a keyed instance is deleted, in milliseconds, by its node's count of its sleep (idle_since of its object): its key is freed, and the next request that carries the key makes a new instance, from the version the app's traffic goes to then; the event sandbox.deleted says reason: idle and the key. A second at the least, and no most. null, which is what it is when nothing is said: never, a keyed instance is the key's until it is deleted. A change holds for the keyed instances the app has: they are looked at again by the app's time as it is, at once, and every minute or as often as the shortest time an app gives, a second apart at the least. An instance with a delete_after_idle_ms of its own (PATCH /v1/sandboxes/{id}) goes by that one instead, longer or shorter: its node deletes it, and its key is freed when the control plane hears of it. A frozen instance (a spend limit; its account's plan, balance or debt) is deleted by neither while it is frozen: once let go of, the sleep it is in counts from then.

        minimum: 1000
    • null
  • egressoptionalobject | null

    Egress rules as a sandbox has them (default, allow, deny, http, inject), which every instance of the app's group has (see App).

    • defaultoptionalallowodenyonull
    • allowoptionalarray | null
    • denyoptionalarray | null
    • httpoptionalarray | null
    • injectoptionalarray | null
    • deny_portsoptionalarray | null

      Ports no connection reaches, TCP or UDP, wherever it goes and whatever else the rules say. A Browsers session's always hold 25, 465, 587 and the ports mining pools listen on (3333, 4444, 5555, 7777, 14444, 45700): the control plane adds them to whatever a change of its rules gives.

  • metadataoptionalobject | null

    Labels: the app's instances carry them. At most 60 keys of up to 64 characters, with values of up to 1024.

  • logsoptional

    What the app's log is to have (see App). Not given: every part at its default.

    Una de estas opciones:

    • AppLogSettings
      • outputoptionalboolean | null

        Whether what the instances print is in the log. false: the machines do not read it; the instances print as before, and an instance keeps the last 16 MiB of it, which is read inside it (GET /v1/sandboxes/{id}/exec/exe-start). What was printed while it was false is not in the log when it is true again.

        default: true
      • keepoptionalboolean | null

        Whether the output is kept for good. false: it is in the log on the machines as before, and counted as not kept (keep_off of GET /v1/apps/{app}/logs/usage); the log says where that begins (logs.keep_off) and ends (logs.keep_on).

        default: true
      • requestsoptional

        Una de estas opciones:

        • object
        • null
    • null
  • tracesoptional

    What the app's requests' traces are to have (see App). Not given: none.

    Una de estas opciones:

    • AppTraceSettings
      • sample_rateoptionalnumber | null

        The share of the app's requests the machine that takes one makes a span for: from when it has read the request's head to when the last byte of its answer has gone, with the request's method, path, status and id, and the instance's wake and an instance made for it as its events. The spans go to the project's destinations that take traces. 0 (the default): none, and the platform takes no part in a request's trace: its traceparent and tracestate reach the instance as they came. Above 0: every request's instance is given a traceparent whose parent is the platform's span, sampled or not, so that the app's own spans (its OpenTelemetry SDK's) are the children of the platform's, and an SDK that goes by its parent's flag samples what the platform samples.

        maximum: 1 · minimum: 0 · default: 0
      • trust_incomingoptionalboolean | null

        Whether a request's own traceparent is taken. true: the platform's span is in the caller's trace, the child of the caller's span, and sampled where the caller says it is (whatever sample_rate is), and the caller's tracestate goes on with it. false (the default): every request begins a trace of its own, sampled at sample_rate, and its span links to the caller's; the caller's tracestate does not reach the instance. Taken, a caller decides what is sampled, and so what is sent to the destinations: for a service open to anybody, leave it false (W3C Trace Context, section 8.1).

        default: false
    • null
  • publicoptionalboolean | null

    Whether the app answers at its default domain on the internet (url). false: it has none, and is reached on its project's network, and at the customers' domains that lead to it. Holds at once. Not given: true.

  • directoptionalboolean | null

    Whether the app answers at a name that does not go through the proxy as well (App.direct). Null: not.

  • checkpointoptional

    Checkpoints of its keyed instances while they run (App.checkpoint). Null: none.

    Una de estas opciones:

    • object
    • null
  • drivesoptionalarray | null

    The Drives its instance mounts read-write (slice D8), kept with each version made after: Drives of the project's, by name or id as written, each looked up when the app is made or changed and when a version is deployed (404 drive_not_found; a volume of another kind is none). A service with a Drive is stateful, as one with a volume of its own is (volumes): one instance, which has its Drives, made on the node that holds them, or on another of their region they are brought to. A Drive has one writer: an app of many instances (scale.max_instances other than 1 or null, scale.min_instances above 1, or keys) is refused, 409 drive_single_writer; read-only mounts of a Drive's snapshots, which many may have, are a sandbox's (mode snapshot, slice D7) and no app's version mounts one yet, and a keyed app gives each key a Drive of its own (key_drives). No two of its volumes and Drives at one path; an instance mounts 8 Drives and volumes of its own at the most (409 too_many_drives). Not given: none.

  • key_drivesoptionalarray | null

    The Drives each key's instance has of its own (slice D8; reports/workers/keyed-upgrade.md, where they were key_volumes), of a keyed app alone (keys; 400 invalid_field otherwise), kept with each version made after. A key's Drives are made when its first instance is made, on the node it is made on and in its region, and mounted before its start command runs (its versions are built with the command held back); every instance the key has after (its first deleted idle, or by its id) mounts the same Drives, by their names; an upgrade of its instance to a version that names other ones mounts them by their names too: one the version names that the key has none of is made then, one it no longer names is left off and kept (POST /v1/apps/{app}/instances/{key}/upgrade); they are deleted with the key (DELETE /v1/apps/{app}/instances/{key}) or with the app. A key's Drive is shown in GET /v1/volumes with its app_id and key, is billed as any Drive, counts toward what the project's Drives hold together and not toward their number, and is not the project's to delete or to mount elsewhere (409 key_drive). Not given: none.

  • volumesoptionalarray | null

    What its instances mount, kept with each version made after: volumes of the project's (POST /v1/volumes), by name or id as written, each looked up when the app is made or changed and when a version is deployed (404 volume_not_found). No two at one path. A service with a volume of its own (one that is not shared) is stateful: it has one instance, which has the volume (scale.max_instances 1 or null, scale.min_instances 0 or 1, no keys: 400 volume_single_instance), on the machine of its volumes, which are on one machine (409 volumes_apart). Its versions are built there with the start command held back; the instance is made at the service's address on the project's network, named as the service is (or as hostname says), mounts the volumes first (a new one given what the image has there, unless nocopy), then starts the command and takes connections once it is up as ready says; every start of it from its disk does the same. It sleeps as any keyed instance does (scale.sleep_after_ms; never with a floor of 1), is woken by a connection on the project's network, and is deleted by its owner alone, or with the app. A version deployed later replaces it: the old one's command is stopped as stop says, the new one is made with the same volumes at the same address, and the traffic is the new one's when it is up; one that does not come up has the version before made again with the volumes as they were left, and the deploy fails with what the new one said (app.deploy_failed). Not given: none.

  • depends_onoptionalarray | null

    The services of the project whose deploy this one's waits for: each once. Holds at once. Not given: none.

  • networkoptional

    How the project's sandboxes find it on its network besides its name. Holds at once: the nodes are told. In a PATCH, merged a key at a time. Not given: no other name, and its names answer its address.

    Una de estas opciones:

    • ServiceNetworkGiven
      • aliasesoptionalarray | null

        Other names: host names of lower-case labels, which may have dots (pg.example.test); not localhost or internal, nothing under either, not beginning with sbx-; each once, and none of them another app's name or alias (409 name_taken). null: none.

      • endpoint_modeoptional

        vip: its names answer its address. dnsrr: they answer the addresses of its instances. null: vip.

        Una de estas opciones:

        • string
        • null
    • null
  • kindoptional

    service: it runs for as long as it is there; job: it runs to its end. Not given: service.

    Una de estas opciones:

    • string
    • null
  • filesoptionalarray | null

    What is written into its instances before the start command runs: no two at one path, and a mebibyte of text in all at the most. A file that holds a secret gives the app the secret: that takes admin, as secrets does, and the secret has to be there. Not given: none.

  • useroptionalstring | null

    The user the start command runs as, by name or number (postgres, 999, 999:999). Not given: the image's own.

    maxLength: 256 · minLength: 1
  • working_diroptionalstring | null

    The directory the start command runs in. Not given: the image's own.

    maxLength: 4096 · pattern: ^/
  • stopoptional

    How an instance is stopped. In a PATCH, merged a key at a time. Not given: the image's own stop signal, else SIGTERM, and ten seconds.

    Una de estas opciones:

    • ServiceStopGiven
      • signaloptionalstring | null

        The signal it is sent first: SIGTERM, SIGINT, SIGQUIT, SIGHUP, SIGKILL, SIGUSR1, SIGUSR2, SIGWINCH or SIGPWR, with or without SIG. null: the image's own, else SIGTERM.

      • grace_msoptionalinteger | null

        How long it is given before it is killed. null: ten seconds.

        maximum: 600000 · minimum: 0
    • null
  • hostnameoptionalstring | null

    The machine name of the instance of a service that has a volume. Not given: the service's name.

    pattern: ^[a-z0-9]([a-z0-9-]{0,61}[a-z0-9])?$
  • runtimeoptional

    The service's runtime settings: limits, sysctls, tmpfs mounts, the resolver, /dev/shm's size. Replaced whole in a PATCH, and null takes them away. Not given: none (open files 1024, raised to 524288 at the most; /dev/shm half of the memory; the guest's own resolver).

    Una de estas opciones:

    • RuntimeGiven
      • ulimitsoptionalobject | null

        Limits by Docker's names (core, cpu, data, fsize, locks, memlock, msgqueue, nice, nofile, nproc, rss, rtprio, rttime, sigpending, stack): a number for both the soft and the hard limit, or {soft, hard}; -1 is no limit. Open files (nofile) from 1 to 1048576, and never -1. Without one: open files 1024, raised to 524288 at the most, as under Docker.

      • sysctlsoptionalobject | null

        Settings of the guest's kernel by name, those Docker lets a container set: net.*, kernel.shm*, kernel.msg*, kernel.sem, fs.mqueue.*. A value of 1 to 1024 bytes on one line. One the guest's kernel does not have fails the build, saying which.

      • tmpfsoptionalarray | null

        File systems in the instance's memory, mounted as it boots.

      • dnsoptional

        Una de estas opciones:

        • RuntimeDnsGiven
        • null
      • shm_size_bytesoptionalinteger | null

        The size of /dev/shm. null: half of the instance's memory.

        maximum: 1099511627776 · minimum: 1048576
    • null

Returns

200

The app after the change, and made_version: the version the change made, null if it made none.

JSON
  • managed_byobject | nullrequired

    What changes it alone, where something does. object database: the database it is under (id, /v1/databases); every other call that would change it is refused, 409 managed_by_database. object sandboxes (id null): the Sandboxes product's, made by the creates of the project's sandboxes (POST /v1/sandboxes): the project's default sandbox app, sbx-sandbox (its image null: each sandbox is made from the base template of the node it goes to, until the project names an image for it, PATCH with image alone, a new version made of that image, of which the creates that name no image are made from then on), or the app of an image a create named (image, as the create named it; sbx-<the image's name>-<12 hex digits>, one version). A version is ready from when it is made, and the app's instances are the project's sandboxes; every call of it but reading it and its versions, naming the default sandbox app's image and deleting it (its versions' templates go with it; the next create of its image, or of none, makes it again) is refused, 409 managed_by_sandboxes. Null: none's. object installation: the installation of a Managed Application it is part of (id, /v1/installations), which changes it alone: every other call that would change it is refused, 409 managed_by_installation.

    • objectdatabaseosandboxesoinstallationrequired
    • idstring | nullrequired
  • publicbooleanrequired

    Whether it answers at its default domain (url).

  • volumesarray deobjectrequired
    • volumestringrequired
    • pathstringrequired
    • read_onlybooleanrequired
    • nocopyoptionalunknown

      There only where it was given as true.

  • drivesarray deobjectrequired

    The Drives its instance mounts (AppCreate.drives): none, [], where it names none.

    • drivestringrequired
    • pathstringrequired
    • modestringrequired
  • key_drivesarray deobjectrequired

    The Drives each key's instance has of its own (AppCreate.key_drives): none, [], where it gives none.

    • namestringrequired
    • pathstringrequired
    • size_bytesinteger | nullrequired
      format: int64
    • nocopyoptionalunknown

      There only where it was given as true.

  • depends_onarray deobjectrequired
    • servicestringrequired
    • conditionstringrequired
  • networkobjectrequired
    • aliasesarray destringrequired
    • endpoint_modestringrequired
  • kindstringrequired
  • filesarray deobjectrequired
    • pathstringrequired
    • modeintegerrequired
    • uidintegerrequired
    • gidintegerrequired
    • contentoptionalstring
    • secretoptionalstring
  • userstring | nullrequired
  • working_dirstring | nullrequired
  • stoprequired

    Una de estas opciones:

    • ServiceStop
      • signalstring | nullrequired
      • grace_msinteger | nullrequired
    • null
  • hostnamestring | nullrequired
  • runtimerequired

    Its runtime settings (AppCreate.runtime). null: none.

    Una de estas opciones:

    • Runtime
      • ulimitsmapa deRuntimeLimitGivenrequired
      • sysctlsmapa destringrequired
      • tmpfsarray deobjectrequired
        • pathstringrequired
        • size_bytesinteger | nullrequired
        • modeintegerrequired
      • dnsobjectrequired
        • serversarray destringrequired
        • searcharray destringrequired
        • optionsarray destringrequired
      • shm_size_bytesinteger | nullrequired
    • null
  • privateobjectrequired

    Where the project's sandboxes on its network find the app: its address there, which it keeps for as long as it is there, renamed or not; and its name under internal. A sandbox of the project on the network finds it by its name, by domain, and by each of its network.aliases with and without .internal.

    • addressstring | nullrequired

      An address of the project's ranges (GET /v1/project). null only where every range of the site is cut.

      pattern: ^10\.(6[4-9]|[7-9][0-9]|1[01][0-9]|12[0-7])(\.(25[0-5]|2[0-4][0-9]|1[0-9][0-9]|[1-9]?[0-9])){2}$
    • domainstringrequired

      <name>.internal.

  • idstringrequired
    pattern: ^app-[0-9a-hjkmnp-tv-z]{16}$
  • objectapprequired
  • namestringrequired
  • imagestring | nullrequired

    What the app's next version is made of, with the seven fields after it: its newest version's, with what was changed since. null: the app has no image, and no version.

  • image_pull_secretstring | nullrequired

    The name of the secret that holds the credentials of the image's registry. No answer has its value. null: none.

  • start_commandrequired

    Una de estas opciones:

    • StartCommand
    • null
  • start_argsrequired

    What follows the image's ENTRYPOINT in the place of its CMD. null: the image's CMD, or the whole of start_command.

    Una de estas opciones:

    • array<string>
    • null
  • portinteger | nullrequired
  • readyobjectrequired

    How to tell that an instance is up, which a version's template is saved after: exactly one of the three. The first two are asked from outside the instance, so the service has to listen on every address (0.0.0.0).

    • httpoptionalstring

      A path, beginning with /: a GET of it on the app's port is answered with a status from 200 to 399.

    • tcpoptionaltrue

      The app's port takes a connection.

    • commandoptional

      A command that succeeds in the instance. It is run by the instance's shell: a list of arguments is written as one line for it.

      Una de estas opciones:

      • string
      • array<string>
  • envmapa destringrequired
  • secretsmapa destringrequired

    Names of secrets. No answer has a value.

  • sizeobjectrequired

    What one instance may use at most. It is billed by what it uses.

    • vcpusintegerrequired
      minimum: 1
    • memory_bytesintegerrequired

      An even number of MiB (a whole number of the 2 MiB pages a node loads a machine's memory in), in bytes. Any other size is 400 invalid_field (param size.memory_bytes): no node would make an instance of it, as none makes a sandbox of it (the node's SandboxLimitsCreate.memory_bytes).

      minimum: 134217728
  • regionsarray destringrequired

    In the order na, eu, ap.

  • scaleobjectrequired
    • min_instancesintegerrequired

      How many instances of the app are kept awake in a region, whether a request comes or not: the floor. The nodes of the region that serve the app share it as they share max_instances (two nodes and a floor of 2: one on each; a floor of 1: one, on the node that is tried first; a floor of 5: three and two), and a node that is away is made up for by the others. A node makes its part when it is told, not at the first request, and neither stops nor deletes those instances for being idle; they are metered as instances that are awake. 0, which is what it is when nothing is said: none. Lowered, the instances beyond it are idle from that moment, and are stopped and deleted as idle instances are. At most max_instances, where that is given, and at most what the account's plan lets an app ask for in a region (caps.app_instances.limit: 5, 50 and 500): more is refused, 403 plan_cap_reached.

      minimum: 0
    • max_instancesinteger | nullrequired

      How many instances the app has at the most in a region: the nodes of the region that serve the app share it. At most what the account's plan lets an app ask for (caps.app_instances.limit: 5 on Free, 500 on Paid): more is refused, 403 plan_cap_reached, and what an app asked for on a plan that let it is kept, and held to the plan's number. null: the plan's number for an app that names none (caps.app_instances.default: 2 on Free, 50 on Paid), or min_instances where that is more. What the nodes are told is no more than the account has room for awake beside everything else of it that is awake, nor less than the app has awake.

      minimum: 1
    • concurrencyinteger | nullrequired

      A cap on what one instance has in hand at a time: the requests the edges sent it, and the connections it has open on its project's network (through its service's address, which the node shares out among the instances connection by connection, or to its own). An instance with that many is full, and another is made for more. null, which is what it is when nothing is said: no cap. A node then adds an instance when the ones it has are near their size in CPU or memory, and an app does not have to guess a number; a service that takes one request at a time gives one.

      maximum: 1000 · minimum: 1
    • sleep_after_msinteger | nullrequired

      The most an instance waits without a request before it stops, in milliseconds. Below it, its node picks the wait by the app's own traffic: the wait that would have cost least over the app's last gaps between a request's end and the next one, a second up and idle against a stop and the resume after it in CPU time, so that an app whose requests come closer together than a stop pays for stays up through them, and one whose requests are far apart stops as soon as its last answer has ended (0: stops at once, always). null, which is what it is when nothing is said: the platform's most, three seconds. An app that said 100, the default before 2026-10-28, says null from then on. One that stays stopped is deleted by its node after delete_after_ms, and the next request that needs an instance gets a new one; the instances min_instances keeps awake are neither stopped nor deleted.

      maximum: 600000 · minimum: 0
    • sleep_notice_msinteger | nullrequired

      How long before an instance stops for being idle the programs connected to its guest's /run/sbx/awake are sent the line sleep N, in milliseconds: they flush what they hold, close their connections, or raise a hold, which puts the stop off. The stop comes that long after, whether they answer or not; where nobody is connected, it is not put off. A keyed instance is told as well. 0 or null: not told (the default).

      maximum: 60000 · minimum: 0
    • sleep_hold_max_msintegerrequired

      How long the holds a program in an instance raises keep it up after its time idle has run out, in milliseconds: a program connects to /run/sbx/awake and sends + before it answers a request whose work goes on after the answer, and - (or closes the connection, or dies) when that work is done; the instance stops sleep_after_ms after the last hold is lowered. Past this time it stops whatever holds it, the programs connected told sleep N cap first where sleep_notice_ms asks for a notice. Held, it runs and is metered as running. 0: holds are not heeded. Null: 30000 (the default).

      maximum: 900000 · minimum: 0
    • delete_after_msinteger | nullrequired

      After how long idle an instance is deleted, counted from where its stop is counted from: the next request that needs one gets a new instance, a clone of the version's template, without what the deleted one had gathered in its memory and on its disk (keep what is to last in a volume or a database). A second to an hour, the platform's time, which is the most: more is refused (400 invalid_field), and an instance that is to stay warm is one of min_instances. null, which is what it is when nothing is said: the hour of the nodes (--app-delete-after-seconds). A change holds for the instances the app has as for the ones made after: each node judges every instance by the app's time as it is, once a second at the least. Not for the app's keyed instances, which keys.delete_after_idle_ms is for.

      maximum: 3600000 · minimum: 1000
    • suspend_after_msoptionalinteger | null

      How long a stopped instance stays paused before its memory is saved and it sleeps for real, in milliseconds: paused, the next request is answered in a millisecond or two and its memory is metered; asleep, it wakes in a few more and only what it keeps is. Its node holds it to the longest it holds any instance paused (ten minutes); the host's memory running short, a slot another machine needs and a sleep somebody asks for put it to sleep sooner. Null (the default): its node's longest. A database's suspend_after_seconds is its app's.

      maximum: 86400000 · minimum: 1000
    • wake_daily_atoptionalstring | null

      A time of day in UTC, HH:MM, at which the app's keyed instances and its stateful service's instance that rest are woken every day, for work of their own that waits for a wake (a database's maintenance, say); the instances the nodes make and delete by the app's requests are woken by requests alone. The wake is no activity: what runs then is, and once that is done the instance rests again. A stateful service's instance is woken besides for the next run of its PostgreSQL server's pg_cron jobs, whatever this says. Null (the default): none. A database's wake_daily_at is its app's.

      pattern: ^([01][0-9]|2[0-3]):[0-5][0-9]$
  • keysrequired

    Una de estas opciones:

    • object
    • null
  • egressobject | nullrequired

    Egress rules as a sandbox has them (default, allow, deny, http, inject), null when there are none; kept as the calls gave them (a PATCH merges them key by key, as a node does). They hold at once and make no version: the nodes that serve the app put them on every instance of its group, the ones there are (one that runs for the connections it has too) and the ones made after; a keyed instance does not have them. The control plane checks only that they are an object of those keys of at most 256 KiB: what the rules say is a node's to check, when it applies them, and an instance whose rules its node cannot read is refused everything (the node's log says why).

    • defaultoptionalallowodenyonull
    • allowoptionalarray | null
    • denyoptionalarray | null
    • httpoptionalarray | null
    • injectoptionalarray | null
    • deny_portsoptionalarray | null

      Ports no connection reaches, TCP or UDP, wherever it goes and whatever else the rules say. A Browsers session's always hold 25, 465, 587 and the ports mining pools listen on (3333, 4444, 5555, 7777, 14444, 45700): the control plane adds them to whatever a change of its rules gives.

  • metadatamapa destringrequired
  • logsobjectrequired

    What the app's log has, each part as it was given or at its default (AppLogSettings).

    • outputbooleanrequired
    • keepbooleanrequired
    • requestsobjectrequired
      • sample_ratenumberrequired
        maximum: 1 · minimum: 0
      • keep_errorsbooleanrequired
  • tracesobjectrequired

    What the app's requests' traces have, each part as it was given or at its default (AppTraceSettings): no span where sample_rate is 0.

    • sample_ratenumberrequired
      maximum: 1 · minimum: 0
    • trust_incomingbooleanrequired
  • urlstring | nullrequired

    Where the app answers: https:// and its default domain, <name>-<the project's short name>.<the zone>. It is the app's from the moment it is made (503 app_not_deployed until its first version is ready) and for as long as the app has its name. null on a control plane that is told no zone for apps (--app-zone).

  • directbooleanrequired

    Whether the app answers at a name that reaches the edges without the platform's proxy as well (direct_url): there a request has no limit of the proxy's on its body or on how long its answer may pause, and nothing but the edge's own limits protects the app.

  • checkpointrequired

    Checkpoints of its keyed instances while they run: every every_seconds (10 unless said, 1 to 600) an instance's memory and disk as they are at one instant go to its node's object store, and when its node's host is lost the instance is taken over on another node from its newest checkpoint with its processes, losing at most what it did since (an instance that sleeps is taken over from its sleep, which loses nothing). Its node pauses it for a few milliseconds for each; what it wrote since the last one is what is sent, compressed, and one that writes more than its node's link for checkpoints takes falls behind instead of filling it (the instance's checkpoint: last_at, behind_seconds, bytes_per_second, stored_bytes). What the store keeps of an instance's newest checkpoint is billed as sleeping state (state) while it runs; nothing for what is sent. Holds at once and makes no version: the instances that are there are given it, those made later are made with it. Not a stateful service's instance: its volume has no copy in the store. An instance's Drives (its own key_drives too) are pinned with its disk at the same instant, and a takeover brings them back as they were then; an instance with a volume that is not a Drive takes none, and its checkpoint.error says so. Null: none (the default).

    Una de estas opciones:

    • object
    • null
  • direct_urlstring | nullrequired

    Where the app answers without the proxy: its default domain's label under the platform's direct zone (https://<name>-<project>.direct.glac.run). Null unless it says direct, and where the platform has no proxy.

  • current_versionstring | nullrequired

    The id of the version the app's traffic goes to: the newest deploy that came up, or what a rollback named. null until a version is ready.

  • deployingstring | nullrequired

    The id of the version that is being deployed and takes the traffic when it is ready.

  • build_failuresarray deobjectrequired

    The nodes that are to serve the version the app's traffic goes to and could not build it when they were asked to after its deploy (a node that joined, one whose template of it is of no use any more, one that came back), one an item, by the node's id; empty where there are none. A node whose build failed is asked again ten minutes after its last try, and is no longer here once it has built the version; the other nodes serve the version meanwhile. Each failure is also an event, app.build_failed, once.

    • node_idstringrequired

      The node that could not build it.

    • version_idstringrequired
      pattern: ^ver-[0-9a-hjkmnp-tv-z]{16}$
    • errorobjectrequired
      • codestringrequired

        The node's code, or this control plane's (secret_not_found, secret_unreadable, invalid_field, secret_host_not_allowed) where the version's secrets could not be read, or may not go to its image's registry.

      • messagestringrequired

        The words of the node, or of the registry, which it passes on without the credentials; or this control plane's, which name the secret.

    • failed_atstringrequired

      When its last try failed.

      format: date-time
  • logs_stateobjectrequired

    How the app's log is kept on each node that keeps one of it, as the node said in its last inventory (every few seconds).

    • nodesarray deobjectrequired

      One item a node, by the node's id; empty where no node keeps a log of the app, or none says (a node of before 2026-10-24).

      • node_idstringrequired
      • for_goodbooleanrequired

        Whether what the app's instances print there now is kept for good: shipped to the object store, kept for the days the plan keeps, and metered. false: it is kept on that node alone, read as ever, not metered, and gone with the node (logs.keep_off in the log). Between the look of the control plane's that stops the account's logs (its plan's are used up and it holds nothing to pay for more) and a node's hearing of it, that node's is still true: what is printed there then is kept for good and metered, and with nothing to pay with it is written off (the statement's written_off_usd).

      • reasonsettingoover_planonullrequired

        Why it is not kept for good: setting, the app's logs.keep; over_plan, its account has used up the logs its plan includes and holds nothing to pay for more. null where it is.

      • sincestring | nullrequired

        Since when it is so, by the node's clock. null: the node noted no time.

        format: date-time
  • created_atstringrequired
    format: date-time
  • updated_atstringrequired
    format: date-time
  • made_versionstring | nullrequired

    The id of the version this call made: a change of what a version is made of, of an app that has an image, makes one, and it is the one the app is deploying when the call answers. null when the call made none (it changed what holds at once, or nothing). deploying alone does not say that: it may be a version another call made.

400

invalid_json, unknown_field, unknown_parameter; invalid_field: a field that is not what its description says (param names it, down to the part: scale.concurrency, ready.http, files[0].path), or a secret that image_pull_secret names and that holds no user name and password (nothing of what it holds is said); invalid_name; invalid_image: not an image reference; invalid_env: a name no environment variable can have, more than 256 KiB, a NUL byte; invalid_reference: a ${{ of env that begins no reference, or, when a version is made, a reference that goes round or refers to a service or a variable that is not there, or to a secret in other text (param is env.<variable>); invalid_metadata; invalid_egress; unknown_region; invalid_idempotency_key; volume_single_instance: a service with a volume of its own (one that is not shared) has one instance, which has the volume, and asks for more: scale.max_instances other than 1, scale.min_instances above 1, or keys (param names which).
JSON
  • errorobjectrequired
    • typestringrequired
    • codestringrequired
    • messagestringrequired
    • paramstring | nullrequired
    • retryablebooleanrequired
    • request_idstringrequired
    • docs_urlstring | nullrequired
  • move_idoptionalstring

    With already_moving: the move of the sandbox that is under way.

  • upgrade_idoptionalstring

    With upgrade_in_progress (the key's upgrade that is not over), upgrade_started (the upgrade that has begun) and rollback_gone (the upgrade whose revert is refused): that upgrade.

    pattern: ^upg-[0-9a-hjkmnp-tv-z]{16}$
  • moveoptionalobject

    With move_over, move_going_back and move_taken_over (POST /internal/moves/{id}/cancel): the move, as it stands.

    • objectstringrequired
    • idstringrequired
      pattern: ^mov-[0-9a-hjkmnp-tv-z]{16}$
    • sandbox_idstringrequired
    • project_idstringrequired
    • fromobjectrequired
      • idstringrequired
      • namestring | nullrequired

        Its name; null for a node that is off the books since.

    • toobjectrequired
      • idstringrequired
      • namestring | nullrequired

        Its name; null for a node that is off the books since.

    • epochintegerrequired

      The epoch the table held the sandbox at when the move began: the old node releases it at it, the new one takes it over one on, the old one takes it back two on, and from the object store three on.

      minimum: 1
    • stepstringrequired

      Where it stands: releasing (the old node is asked to release it; the table has it there), taking_over (the new node is asked to take it over; the table has it there), reclaiming (the new node refused it, and the old one is asked to take it back), restoring (the old one kept nothing of it, and takes it over from the object store), done (on the new node), failed (where it was, or back there, or where a failure no time clears or an operator's cancel found it: error says why).

    • whystringrequired

      What made it: the operator's move, an evacuation, a recovery.

    • wakebooleanrequired

      Whether the new node wakes it: as the call said, or as it was (it ran) when the move began.

    • from_storebooleanrequired

      The new node takes it from the object store alone.

    • releaseobject | nullrequired

      What the old node's release answered (sandboxd/openapi/openapi.json, releaseSandbox), without the release's key; null before it answered.

    • takeoverobject | nullrequired

      What the takeover that ended it answered (takeoverSandbox of the node's document): where its disk and memory came from (source), how long each took (ms), memory; null before.

    • errorrequired

      Why it failed, of one that failed (the new node's refusal, where it was taken back to the old node); null, of one that is done; of one under way, what the step it is at was last told where it is asked again.

      Una de estas opciones:

      • object
      • null
    • attemptsintegerrequired

      How many times the step it is at was asked without an answer that settles it.

      minimum: 0
    • next_attempt_atstring | nullrequired

      When that step is asked again; null where it is not waiting.

      format: date-time
    • batchstring | nullrequired

      The evacuation (evc-…) or recovery (rcv-…) it is of.

    • created_atstringrequired
      format: date-time
    • updated_atstringrequired
      format: date-time
    • ended_atstring | nullrequired
      format: date-time
  • lease_untiloptionalstring

    With lease_held: until when the node may still hold the lease it was granted last, and the time after it for the node to have seen it run out (recoverNode is refused until then).

    format: date-time
  • activeoptionalobject | null

    With standby: the active copy, as the standby last read the lease; null while no copy holds it.

    • namestringrequired
    • urlstringrequired
    • termintegerrequired
      minimum: 1
  • allowed_hostsoptionalarray | null

    With secret_host_not_allowed: the hosts the secret may go to, as the secret has them; null where it was given none (the node's SecretHosts).

  • problemsoptionalarray deobject

    With invalid_definition: every problem of the definition, each where it is and what it is.

    • atstringrequired

      Where in the definition, a dotted path (services.web.port); empty for the whole of it.

    • messagestringrequired
  • reportoptionalobject

    With compose_refused: the import's report, which says what is refused, why, and what to change.

    • objectcompose_importrequired
    • namestring | nullrequired

      The model's name: each service's label compose.project. Null: it has none.

    • dry_runbooleanrequired
    • strictbooleanrequired
    • profilesarray destringrequired
    • treatmentsobjectrequired

      How many items of the report have each treatment.

      • doneintegerrequired
        minimum: 0
      • adaptedintegerrequired
        minimum: 0
      • ignoredintegerrequired
        minimum: 0
      • refusedintegerrequired
        minimum: 0
    • itemsarray deobjectrequired

      The model's top-level elements: name, each volume, network, config and secret, extensions.

      • atstringrequired

        Where in the model: of a service, its attribute (ports[0], depends_on.db, environment.KEY), empty for the service as a whole; of the top level, the element (volumes.db-data).

      • treatmentdoneoadaptedoignoredorefusedrequired

        done: it means the same on the platform; adapted: the platform does the same another way; ignored: it matters to a local run alone (refused with strict); refused: the platform cannot do it, and doing without it would let the program go wrong quietly.

      • whystringrequired
      • changestring | nullrequired

        Of one that is refused: what to change. Null otherwise.

    • servicesarray deobjectrequired

      The services the import takes, in the model's order.

      • namestringrequired

        The service's name, which is its app's.

      • app_idstring | nullrequired

        Its app. Null: a dry run of a service the project does not have.

        pattern: ^app-
      • statuscreatedochangedounchangedrequired

        What the import does (or, in a dry run, would do) to its app: makes it, writes what changed, or nothing.

      • kindserviceojobrequired

        job: a service of the model waits for it to complete, and it runs to its end in each deploy that needs it.

      • statefulbooleanrequired

        It mounts a volume of its own: one instance, which has the volume.

      • versionbooleanrequired

        Whether the import's deploy makes a version of it: what its version would be made of, its references resolved, is not what its current version is.

      • itemsarray deobjectrequired
        • atstringrequired

          Where in the model: of a service, its attribute (ports[0], depends_on.db, environment.KEY), empty for the service as a whole; of the top level, the element (volumes.db-data).

        • treatmentdoneoadaptedoignoredorefusedrequired

          done: it means the same on the platform; adapted: the platform does the same another way; ignored: it matters to a local run alone (refused with strict); refused: the platform cannot do it, and doing without it would let the program go wrong quietly.

        • whystringrequired
        • changestring | nullrequired

          Of one that is refused: what to change. Null otherwise.

      • appobjectrequired

        The fields of the service's app as the import writes them: what its versions are made of, and what holds at once.

        • imagestring | nullrequired

          What the app's next version is made of, with the seven fields after it: its newest version's, with what was changed since. null: the app has no image, and no version.

        • image_pull_secretstring | nullrequired

          The name of the secret that holds the credentials of the image's registry. No answer has its value. null: none.

        • start_commandrequired

          Una de estas opciones:

          • StartCommand
          • null
        • start_argsrequired

          What follows the image's ENTRYPOINT in the place of its CMD. null: the image's CMD, or the whole of start_command.

          Una de estas opciones:

          • array<string>
          • null
        • portinteger | nullrequired
        • readyobjectrequired

          How to tell that an instance is up, which a version's template is saved after: exactly one of the three. The first two are asked from outside the instance, so the service has to listen on every address (0.0.0.0).

          • httpoptionalstring

            A path, beginning with /: a GET of it on the app's port is answered with a status from 200 to 399.

          • tcpoptionaltrue

            The app's port takes a connection.

          • commandoptional

            A command that succeeds in the instance. It is run by the instance's shell: a list of arguments is written as one line for it.

            Una de estas opciones:

            • string
            • array<string>
        • envmapa destringrequired
        • secretsmapa destringrequired

          Names of secrets. No answer has a value.

        • sizeobjectrequired

          What one instance may use at most. It is billed by what it uses.

          • vcpusintegerrequired
            minimum: 1
          • memory_bytesintegerrequired

            An even number of MiB (a whole number of the 2 MiB pages a node loads a machine's memory in), in bytes. Any other size is 400 invalid_field (param size.memory_bytes): no node would make an instance of it, as none makes a sandbox of it (the node's SandboxLimitsCreate.memory_bytes).

            minimum: 134217728
        • volumesarray deobjectrequired
          • volumestringrequired
          • pathstringrequired
          • read_onlybooleanrequired
          • nocopyoptionalunknown

            There only where it was given as true.

        • kindstringrequired
        • filesarray deobjectrequired
          • pathstringrequired
          • modeintegerrequired
          • uidintegerrequired
          • gidintegerrequired
          • contentoptionalstring
          • secretoptionalstring
        • userstring | nullrequired
        • working_dirstring | nullrequired
        • stoprequired

          Una de estas opciones:

          • ServiceStop
            • signalstring | nullrequired
            • grace_msinteger | nullrequired
          • null
        • hostnamestring | nullrequired
        • regionsarray destringrequired

          In the order na, eu, ap.

        • scaleobjectrequired
          • min_instancesintegerrequired

            How many instances of the app are kept awake in a region, whether a request comes or not: the floor. The nodes of the region that serve the app share it as they share max_instances (two nodes and a floor of 2: one on each; a floor of 1: one, on the node that is tried first; a floor of 5: three and two), and a node that is away is made up for by the others. A node makes its part when it is told, not at the first request, and neither stops nor deletes those instances for being idle; they are metered as instances that are awake. 0, which is what it is when nothing is said: none. Lowered, the instances beyond it are idle from that moment, and are stopped and deleted as idle instances are. At most max_instances, where that is given, and at most what the account's plan lets an app ask for in a region (caps.app_instances.limit: 5, 50 and 500): more is refused, 403 plan_cap_reached.

            minimum: 0
          • max_instancesinteger | nullrequired

            How many instances the app has at the most in a region: the nodes of the region that serve the app share it. At most what the account's plan lets an app ask for (caps.app_instances.limit: 5 on Free, 500 on Paid): more is refused, 403 plan_cap_reached, and what an app asked for on a plan that let it is kept, and held to the plan's number. null: the plan's number for an app that names none (caps.app_instances.default: 2 on Free, 50 on Paid), or min_instances where that is more. What the nodes are told is no more than the account has room for awake beside everything else of it that is awake, nor less than the app has awake.

            minimum: 1
          • concurrencyinteger | nullrequired

            A cap on what one instance has in hand at a time: the requests the edges sent it, and the connections it has open on its project's network (through its service's address, which the node shares out among the instances connection by connection, or to its own). An instance with that many is full, and another is made for more. null, which is what it is when nothing is said: no cap. A node then adds an instance when the ones it has are near their size in CPU or memory, and an app does not have to guess a number; a service that takes one request at a time gives one.

            maximum: 1000 · minimum: 1
          • sleep_after_msinteger | nullrequired

            The most an instance waits without a request before it stops, in milliseconds. Below it, its node picks the wait by the app's own traffic: the wait that would have cost least over the app's last gaps between a request's end and the next one, a second up and idle against a stop and the resume after it in CPU time, so that an app whose requests come closer together than a stop pays for stays up through them, and one whose requests are far apart stops as soon as its last answer has ended (0: stops at once, always). null, which is what it is when nothing is said: the platform's most, three seconds. An app that said 100, the default before 2026-10-28, says null from then on. One that stays stopped is deleted by its node after delete_after_ms, and the next request that needs an instance gets a new one; the instances min_instances keeps awake are neither stopped nor deleted.

            maximum: 600000 · minimum: 0
          • sleep_notice_msinteger | nullrequired

            How long before an instance stops for being idle the programs connected to its guest's /run/sbx/awake are sent the line sleep N, in milliseconds: they flush what they hold, close their connections, or raise a hold, which puts the stop off. The stop comes that long after, whether they answer or not; where nobody is connected, it is not put off. A keyed instance is told as well. 0 or null: not told (the default).

            maximum: 60000 · minimum: 0
          • sleep_hold_max_msintegerrequired

            How long the holds a program in an instance raises keep it up after its time idle has run out, in milliseconds: a program connects to /run/sbx/awake and sends + before it answers a request whose work goes on after the answer, and - (or closes the connection, or dies) when that work is done; the instance stops sleep_after_ms after the last hold is lowered. Past this time it stops whatever holds it, the programs connected told sleep N cap first where sleep_notice_ms asks for a notice. Held, it runs and is metered as running. 0: holds are not heeded. Null: 30000 (the default).

            maximum: 900000 · minimum: 0
          • delete_after_msinteger | nullrequired

            After how long idle an instance is deleted, counted from where its stop is counted from: the next request that needs one gets a new instance, a clone of the version's template, without what the deleted one had gathered in its memory and on its disk (keep what is to last in a volume or a database). A second to an hour, the platform's time, which is the most: more is refused (400 invalid_field), and an instance that is to stay warm is one of min_instances. null, which is what it is when nothing is said: the hour of the nodes (--app-delete-after-seconds). A change holds for the instances the app has as for the ones made after: each node judges every instance by the app's time as it is, once a second at the least. Not for the app's keyed instances, which keys.delete_after_idle_ms is for.

            maximum: 3600000 · minimum: 1000
          • suspend_after_msoptionalinteger | null

            How long a stopped instance stays paused before its memory is saved and it sleeps for real, in milliseconds: paused, the next request is answered in a millisecond or two and its memory is metered; asleep, it wakes in a few more and only what it keeps is. Its node holds it to the longest it holds any instance paused (ten minutes); the host's memory running short, a slot another machine needs and a sleep somebody asks for put it to sleep sooner. Null (the default): its node's longest. A database's suspend_after_seconds is its app's.

            maximum: 86400000 · minimum: 1000
          • wake_daily_atoptionalstring | null

            A time of day in UTC, HH:MM, at which the app's keyed instances and its stateful service's instance that rest are woken every day, for work of their own that waits for a wake (a database's maintenance, say); the instances the nodes make and delete by the app's requests are woken by requests alone. The wake is no activity: what runs then is, and once that is done the instance rests again. A stateful service's instance is woken besides for the next run of its PostgreSQL server's pg_cron jobs, whatever this says. Null (the default): none. A database's wake_daily_at is its app's.

            pattern: ^([01][0-9]|2[0-3]):[0-5][0-9]$
        • keysrequired

          Una de estas opciones:

          • object
          • null
        • egressobject | nullrequired

          Egress rules as a sandbox has them (default, allow, deny, http, inject), null when there are none; kept as the calls gave them (a PATCH merges them key by key, as a node does). They hold at once and make no version: the nodes that serve the app put them on every instance of its group, the ones there are (one that runs for the connections it has too) and the ones made after; a keyed instance does not have them. The control plane checks only that they are an object of those keys of at most 256 KiB: what the rules say is a node's to check, when it applies them, and an instance whose rules its node cannot read is refused everything (the node's log says why).

          • defaultoptionalallowodenyonull
          • allowoptionalarray | null
          • denyoptionalarray | null
          • httpoptionalarray | null
          • injectoptionalarray | null
          • deny_portsoptionalarray | null

            Ports no connection reaches, TCP or UDP, wherever it goes and whatever else the rules say. A Browsers session's always hold 25, 465, 587 and the ports mining pools listen on (3333, 4444, 5555, 7777, 14444, 45700): the control plane adds them to whatever a change of its rules gives.

        • metadatamapa destringrequired
        • logsobjectrequired

          What the app's log has, each part as it was given or at its default (AppLogSettings).

          • outputbooleanrequired
          • keepbooleanrequired
          • requestsobjectrequired
            • sample_ratenumberrequired
              maximum: 1 · minimum: 0
            • keep_errorsbooleanrequired
        • publicbooleanrequired

          Whether it answers at its default domain (url).

        • depends_onarray deobjectrequired
          • servicestringrequired
          • conditionstringrequired
        • networkobjectrequired
          • aliasesarray destringrequired
          • endpoint_modestringrequired
        • drivesoptionalarray deobject

          The Drives its instance mounts (AppCreate.drives): none, [], where it names none.

          • drivestringrequired
          • pathstringrequired
          • modestringrequired
        • key_drivesoptionalarray deobject

          The Drives each key's instance has of its own (AppCreate.key_drives): none, [], where it gives none.

          • namestringrequired
          • pathstringrequired
          • size_bytesinteger | nullrequired
            format: int64
          • nocopyoptionalunknown

            There only where it was given as true.

        • runtimeoptional

          Its runtime settings (AppCreate.runtime). null: none.

          Una de estas opciones:

          • Runtime
            • ulimitsmapa deRuntimeLimitGivenrequired
            • sysctlsmapa destringrequired
            • tmpfsarray deobjectrequired
              • pathstringrequired
              • size_bytesinteger | nullrequired
              • modeintegerrequired
            • dnsobjectrequired
              • serversarray destringrequired
              • searcharray destringrequired
              • optionsarray destringrequired
            • shm_size_bytesinteger | nullrequired
          • null
        • tracesoptionalobject

          What the app's requests' traces have, each part as it was given or at its default (AppTraceSettings): no span where sample_rate is 0.

          • sample_ratenumberrequired
            maximum: 1 · minimum: 0
          • trust_incomingbooleanrequired
    • volumesarray deobjectrequired

      The volumes of the project the services mount.

      • modelstring | nullrequired

        Its name in the model. Null: made for a path a stateful service mounts with no volume named.

      • namestringrequired

        Its name in the project.

      • idstring | nullrequired

        Null: not made yet (to_make).

        pattern: ^vol-
      • sharedbooleanrequired

        Several services mount it, or one with several instances: every instance of each mounts it at once.

      • statusto_makeomadeothereoexternalrequired

        to_make: the import makes it (a dry run says so of what it would make); made: this import made it; there: an earlier import made it, or a volume of its name was there; external: the model names it external, and the project has it.

      • used_byarray destringrequired
    • secretsarray deobjectrequired

      The secrets of the model that services are granted.

      • modelstringrequired

        Its name in the model.

      • namestringrequired

        The project's secret that holds it: the model's name, with each character a secret's name does not take written as an underscore.

      • therebooleanrequired

        Whether the project has it. An import (not a dry run) with one that is not there is refused 404 secret_not_found.

      • used_byarray destringrequired
    • left_outarray deobjectrequired

      The services of profiles that are not turned on, which the import does not touch.

      • namestringrequired
      • profilesarray destringrequired
    • absentarray destringrequired

      The project's services that the model does not have: kept as they are.

    • deploystring | nullrequired

      The deploy the import made (GET /v1/deploys/{id}). Null: a dry run, or nothing would get a version.

      pattern: ^dpl-
  • accountsoptionalarray deobject

    With sole_owner: the accounts the user is the only owner of.

    • idstringrequired
      pattern: ^acct-[0-9a-hjkmnp-tv-z]{16}$
    • namestringrequired

401

missing_api_key, invalid_api_key, expired_api_key, invalid_token.
JSON
  • errorobjectrequired
    • typestringrequired
    • codestringrequired
    • messagestringrequired
    • paramstring | nullrequired
    • retryablebooleanrequired
    • request_idstringrequired
    • docs_urlstring | nullrequired
  • move_idoptionalstring

    With already_moving: the move of the sandbox that is under way.

  • upgrade_idoptionalstring

    With upgrade_in_progress (the key's upgrade that is not over), upgrade_started (the upgrade that has begun) and rollback_gone (the upgrade whose revert is refused): that upgrade.

    pattern: ^upg-[0-9a-hjkmnp-tv-z]{16}$
  • moveoptionalobject

    With move_over, move_going_back and move_taken_over (POST /internal/moves/{id}/cancel): the move, as it stands.

    • objectstringrequired
    • idstringrequired
      pattern: ^mov-[0-9a-hjkmnp-tv-z]{16}$
    • sandbox_idstringrequired
    • project_idstringrequired
    • fromobjectrequired
      • idstringrequired
      • namestring | nullrequired

        Its name; null for a node that is off the books since.

    • toobjectrequired
      • idstringrequired
      • namestring | nullrequired

        Its name; null for a node that is off the books since.

    • epochintegerrequired

      The epoch the table held the sandbox at when the move began: the old node releases it at it, the new one takes it over one on, the old one takes it back two on, and from the object store three on.

      minimum: 1
    • stepstringrequired

      Where it stands: releasing (the old node is asked to release it; the table has it there), taking_over (the new node is asked to take it over; the table has it there), reclaiming (the new node refused it, and the old one is asked to take it back), restoring (the old one kept nothing of it, and takes it over from the object store), done (on the new node), failed (where it was, or back there, or where a failure no time clears or an operator's cancel found it: error says why).

    • whystringrequired

      What made it: the operator's move, an evacuation, a recovery.

    • wakebooleanrequired

      Whether the new node wakes it: as the call said, or as it was (it ran) when the move began.

    • from_storebooleanrequired

      The new node takes it from the object store alone.

    • releaseobject | nullrequired

      What the old node's release answered (sandboxd/openapi/openapi.json, releaseSandbox), without the release's key; null before it answered.

    • takeoverobject | nullrequired

      What the takeover that ended it answered (takeoverSandbox of the node's document): where its disk and memory came from (source), how long each took (ms), memory; null before.

    • errorrequired

      Why it failed, of one that failed (the new node's refusal, where it was taken back to the old node); null, of one that is done; of one under way, what the step it is at was last told where it is asked again.

      Una de estas opciones:

      • object
      • null
    • attemptsintegerrequired

      How many times the step it is at was asked without an answer that settles it.

      minimum: 0
    • next_attempt_atstring | nullrequired

      When that step is asked again; null where it is not waiting.

      format: date-time
    • batchstring | nullrequired

      The evacuation (evc-…) or recovery (rcv-…) it is of.

    • created_atstringrequired
      format: date-time
    • updated_atstringrequired
      format: date-time
    • ended_atstring | nullrequired
      format: date-time
  • lease_untiloptionalstring

    With lease_held: until when the node may still hold the lease it was granted last, and the time after it for the node to have seen it run out (recoverNode is refused until then).

    format: date-time
  • activeoptionalobject | null

    With standby: the active copy, as the standby last read the lease; null while no copy holds it.

    • namestringrequired
    • urlstringrequired
    • termintegerrequired
      minimum: 1
  • allowed_hostsoptionalarray | null

    With secret_host_not_allowed: the hosts the secret may go to, as the secret has them; null where it was given none (the node's SecretHosts).

  • problemsoptionalarray deobject

    With invalid_definition: every problem of the definition, each where it is and what it is.

    • atstringrequired

      Where in the definition, a dotted path (services.web.port); empty for the whole of it.

    • messagestringrequired
  • reportoptionalobject

    With compose_refused: the import's report, which says what is refused, why, and what to change.

    • objectcompose_importrequired
    • namestring | nullrequired

      The model's name: each service's label compose.project. Null: it has none.

    • dry_runbooleanrequired
    • strictbooleanrequired
    • profilesarray destringrequired
    • treatmentsobjectrequired

      How many items of the report have each treatment.

      • doneintegerrequired
        minimum: 0
      • adaptedintegerrequired
        minimum: 0
      • ignoredintegerrequired
        minimum: 0
      • refusedintegerrequired
        minimum: 0
    • itemsarray deobjectrequired

      The model's top-level elements: name, each volume, network, config and secret, extensions.

      • atstringrequired

        Where in the model: of a service, its attribute (ports[0], depends_on.db, environment.KEY), empty for the service as a whole; of the top level, the element (volumes.db-data).

      • treatmentdoneoadaptedoignoredorefusedrequired

        done: it means the same on the platform; adapted: the platform does the same another way; ignored: it matters to a local run alone (refused with strict); refused: the platform cannot do it, and doing without it would let the program go wrong quietly.

      • whystringrequired
      • changestring | nullrequired

        Of one that is refused: what to change. Null otherwise.

    • servicesarray deobjectrequired

      The services the import takes, in the model's order.

      • namestringrequired

        The service's name, which is its app's.

      • app_idstring | nullrequired

        Its app. Null: a dry run of a service the project does not have.

        pattern: ^app-
      • statuscreatedochangedounchangedrequired

        What the import does (or, in a dry run, would do) to its app: makes it, writes what changed, or nothing.

      • kindserviceojobrequired

        job: a service of the model waits for it to complete, and it runs to its end in each deploy that needs it.

      • statefulbooleanrequired

        It mounts a volume of its own: one instance, which has the volume.

      • versionbooleanrequired

        Whether the import's deploy makes a version of it: what its version would be made of, its references resolved, is not what its current version is.

      • itemsarray deobjectrequired
        • atstringrequired

          Where in the model: of a service, its attribute (ports[0], depends_on.db, environment.KEY), empty for the service as a whole; of the top level, the element (volumes.db-data).

        • treatmentdoneoadaptedoignoredorefusedrequired

          done: it means the same on the platform; adapted: the platform does the same another way; ignored: it matters to a local run alone (refused with strict); refused: the platform cannot do it, and doing without it would let the program go wrong quietly.

        • whystringrequired
        • changestring | nullrequired

          Of one that is refused: what to change. Null otherwise.

      • appobjectrequired

        The fields of the service's app as the import writes them: what its versions are made of, and what holds at once.

        • imagestring | nullrequired

          What the app's next version is made of, with the seven fields after it: its newest version's, with what was changed since. null: the app has no image, and no version.

        • image_pull_secretstring | nullrequired

          The name of the secret that holds the credentials of the image's registry. No answer has its value. null: none.

        • start_commandrequired

          Una de estas opciones:

          • StartCommand
          • null
        • start_argsrequired

          What follows the image's ENTRYPOINT in the place of its CMD. null: the image's CMD, or the whole of start_command.

          Una de estas opciones:

          • array<string>
          • null
        • portinteger | nullrequired
        • readyobjectrequired

          How to tell that an instance is up, which a version's template is saved after: exactly one of the three. The first two are asked from outside the instance, so the service has to listen on every address (0.0.0.0).

          • httpoptionalstring

            A path, beginning with /: a GET of it on the app's port is answered with a status from 200 to 399.

          • tcpoptionaltrue

            The app's port takes a connection.

          • commandoptional

            A command that succeeds in the instance. It is run by the instance's shell: a list of arguments is written as one line for it.

            Una de estas opciones:

            • string
            • array<string>
        • envmapa destringrequired
        • secretsmapa destringrequired

          Names of secrets. No answer has a value.

        • sizeobjectrequired

          What one instance may use at most. It is billed by what it uses.

          • vcpusintegerrequired
            minimum: 1
          • memory_bytesintegerrequired

            An even number of MiB (a whole number of the 2 MiB pages a node loads a machine's memory in), in bytes. Any other size is 400 invalid_field (param size.memory_bytes): no node would make an instance of it, as none makes a sandbox of it (the node's SandboxLimitsCreate.memory_bytes).

            minimum: 134217728
        • volumesarray deobjectrequired
          • volumestringrequired
          • pathstringrequired
          • read_onlybooleanrequired
          • nocopyoptionalunknown

            There only where it was given as true.

        • kindstringrequired
        • filesarray deobjectrequired
          • pathstringrequired
          • modeintegerrequired
          • uidintegerrequired
          • gidintegerrequired
          • contentoptionalstring
          • secretoptionalstring
        • userstring | nullrequired
        • working_dirstring | nullrequired
        • stoprequired

          Una de estas opciones:

          • ServiceStop
            • signalstring | nullrequired
            • grace_msinteger | nullrequired
          • null
        • hostnamestring | nullrequired
        • regionsarray destringrequired

          In the order na, eu, ap.

        • scaleobjectrequired
          • min_instancesintegerrequired

            How many instances of the app are kept awake in a region, whether a request comes or not: the floor. The nodes of the region that serve the app share it as they share max_instances (two nodes and a floor of 2: one on each; a floor of 1: one, on the node that is tried first; a floor of 5: three and two), and a node that is away is made up for by the others. A node makes its part when it is told, not at the first request, and neither stops nor deletes those instances for being idle; they are metered as instances that are awake. 0, which is what it is when nothing is said: none. Lowered, the instances beyond it are idle from that moment, and are stopped and deleted as idle instances are. At most max_instances, where that is given, and at most what the account's plan lets an app ask for in a region (caps.app_instances.limit: 5, 50 and 500): more is refused, 403 plan_cap_reached.

            minimum: 0
          • max_instancesinteger | nullrequired

            How many instances the app has at the most in a region: the nodes of the region that serve the app share it. At most what the account's plan lets an app ask for (caps.app_instances.limit: 5 on Free, 500 on Paid): more is refused, 403 plan_cap_reached, and what an app asked for on a plan that let it is kept, and held to the plan's number. null: the plan's number for an app that names none (caps.app_instances.default: 2 on Free, 50 on Paid), or min_instances where that is more. What the nodes are told is no more than the account has room for awake beside everything else of it that is awake, nor less than the app has awake.

            minimum: 1
          • concurrencyinteger | nullrequired

            A cap on what one instance has in hand at a time: the requests the edges sent it, and the connections it has open on its project's network (through its service's address, which the node shares out among the instances connection by connection, or to its own). An instance with that many is full, and another is made for more. null, which is what it is when nothing is said: no cap. A node then adds an instance when the ones it has are near their size in CPU or memory, and an app does not have to guess a number; a service that takes one request at a time gives one.

            maximum: 1000 · minimum: 1
          • sleep_after_msinteger | nullrequired

            The most an instance waits without a request before it stops, in milliseconds. Below it, its node picks the wait by the app's own traffic: the wait that would have cost least over the app's last gaps between a request's end and the next one, a second up and idle against a stop and the resume after it in CPU time, so that an app whose requests come closer together than a stop pays for stays up through them, and one whose requests are far apart stops as soon as its last answer has ended (0: stops at once, always). null, which is what it is when nothing is said: the platform's most, three seconds. An app that said 100, the default before 2026-10-28, says null from then on. One that stays stopped is deleted by its node after delete_after_ms, and the next request that needs an instance gets a new one; the instances min_instances keeps awake are neither stopped nor deleted.

            maximum: 600000 · minimum: 0
          • sleep_notice_msinteger | nullrequired

            How long before an instance stops for being idle the programs connected to its guest's /run/sbx/awake are sent the line sleep N, in milliseconds: they flush what they hold, close their connections, or raise a hold, which puts the stop off. The stop comes that long after, whether they answer or not; where nobody is connected, it is not put off. A keyed instance is told as well. 0 or null: not told (the default).

            maximum: 60000 · minimum: 0
          • sleep_hold_max_msintegerrequired

            How long the holds a program in an instance raises keep it up after its time idle has run out, in milliseconds: a program connects to /run/sbx/awake and sends + before it answers a request whose work goes on after the answer, and - (or closes the connection, or dies) when that work is done; the instance stops sleep_after_ms after the last hold is lowered. Past this time it stops whatever holds it, the programs connected told sleep N cap first where sleep_notice_ms asks for a notice. Held, it runs and is metered as running. 0: holds are not heeded. Null: 30000 (the default).

            maximum: 900000 · minimum: 0
          • delete_after_msinteger | nullrequired

            After how long idle an instance is deleted, counted from where its stop is counted from: the next request that needs one gets a new instance, a clone of the version's template, without what the deleted one had gathered in its memory and on its disk (keep what is to last in a volume or a database). A second to an hour, the platform's time, which is the most: more is refused (400 invalid_field), and an instance that is to stay warm is one of min_instances. null, which is what it is when nothing is said: the hour of the nodes (--app-delete-after-seconds). A change holds for the instances the app has as for the ones made after: each node judges every instance by the app's time as it is, once a second at the least. Not for the app's keyed instances, which keys.delete_after_idle_ms is for.

            maximum: 3600000 · minimum: 1000
          • suspend_after_msoptionalinteger | null

            How long a stopped instance stays paused before its memory is saved and it sleeps for real, in milliseconds: paused, the next request is answered in a millisecond or two and its memory is metered; asleep, it wakes in a few more and only what it keeps is. Its node holds it to the longest it holds any instance paused (ten minutes); the host's memory running short, a slot another machine needs and a sleep somebody asks for put it to sleep sooner. Null (the default): its node's longest. A database's suspend_after_seconds is its app's.

            maximum: 86400000 · minimum: 1000
          • wake_daily_atoptionalstring | null

            A time of day in UTC, HH:MM, at which the app's keyed instances and its stateful service's instance that rest are woken every day, for work of their own that waits for a wake (a database's maintenance, say); the instances the nodes make and delete by the app's requests are woken by requests alone. The wake is no activity: what runs then is, and once that is done the instance rests again. A stateful service's instance is woken besides for the next run of its PostgreSQL server's pg_cron jobs, whatever this says. Null (the default): none. A database's wake_daily_at is its app's.

            pattern: ^([01][0-9]|2[0-3]):[0-5][0-9]$
        • keysrequired

          Una de estas opciones:

          • object
          • null
        • egressobject | nullrequired

          Egress rules as a sandbox has them (default, allow, deny, http, inject), null when there are none; kept as the calls gave them (a PATCH merges them key by key, as a node does). They hold at once and make no version: the nodes that serve the app put them on every instance of its group, the ones there are (one that runs for the connections it has too) and the ones made after; a keyed instance does not have them. The control plane checks only that they are an object of those keys of at most 256 KiB: what the rules say is a node's to check, when it applies them, and an instance whose rules its node cannot read is refused everything (the node's log says why).

          • defaultoptionalallowodenyonull
          • allowoptionalarray | null
          • denyoptionalarray | null
          • httpoptionalarray | null
          • injectoptionalarray | null
          • deny_portsoptionalarray | null

            Ports no connection reaches, TCP or UDP, wherever it goes and whatever else the rules say. A Browsers session's always hold 25, 465, 587 and the ports mining pools listen on (3333, 4444, 5555, 7777, 14444, 45700): the control plane adds them to whatever a change of its rules gives.

        • metadatamapa destringrequired
        • logsobjectrequired

          What the app's log has, each part as it was given or at its default (AppLogSettings).

          • outputbooleanrequired
          • keepbooleanrequired
          • requestsobjectrequired
            • sample_ratenumberrequired
              maximum: 1 · minimum: 0
            • keep_errorsbooleanrequired
        • publicbooleanrequired

          Whether it answers at its default domain (url).

        • depends_onarray deobjectrequired
          • servicestringrequired
          • conditionstringrequired
        • networkobjectrequired
          • aliasesarray destringrequired
          • endpoint_modestringrequired
        • drivesoptionalarray deobject

          The Drives its instance mounts (AppCreate.drives): none, [], where it names none.

          • drivestringrequired
          • pathstringrequired
          • modestringrequired
        • key_drivesoptionalarray deobject

          The Drives each key's instance has of its own (AppCreate.key_drives): none, [], where it gives none.

          • namestringrequired
          • pathstringrequired
          • size_bytesinteger | nullrequired
            format: int64
          • nocopyoptionalunknown

            There only where it was given as true.

        • runtimeoptional

          Its runtime settings (AppCreate.runtime). null: none.

          Una de estas opciones:

          • Runtime
            • ulimitsmapa deRuntimeLimitGivenrequired
            • sysctlsmapa destringrequired
            • tmpfsarray deobjectrequired
              • pathstringrequired
              • size_bytesinteger | nullrequired
              • modeintegerrequired
            • dnsobjectrequired
              • serversarray destringrequired
              • searcharray destringrequired
              • optionsarray destringrequired
            • shm_size_bytesinteger | nullrequired
          • null
        • tracesoptionalobject

          What the app's requests' traces have, each part as it was given or at its default (AppTraceSettings): no span where sample_rate is 0.

          • sample_ratenumberrequired
            maximum: 1 · minimum: 0
          • trust_incomingbooleanrequired
    • volumesarray deobjectrequired

      The volumes of the project the services mount.

      • modelstring | nullrequired

        Its name in the model. Null: made for a path a stateful service mounts with no volume named.

      • namestringrequired

        Its name in the project.

      • idstring | nullrequired

        Null: not made yet (to_make).

        pattern: ^vol-
      • sharedbooleanrequired

        Several services mount it, or one with several instances: every instance of each mounts it at once.

      • statusto_makeomadeothereoexternalrequired

        to_make: the import makes it (a dry run says so of what it would make); made: this import made it; there: an earlier import made it, or a volume of its name was there; external: the model names it external, and the project has it.

      • used_byarray destringrequired
    • secretsarray deobjectrequired

      The secrets of the model that services are granted.

      • modelstringrequired

        Its name in the model.

      • namestringrequired

        The project's secret that holds it: the model's name, with each character a secret's name does not take written as an underscore.

      • therebooleanrequired

        Whether the project has it. An import (not a dry run) with one that is not there is refused 404 secret_not_found.

      • used_byarray destringrequired
    • left_outarray deobjectrequired

      The services of profiles that are not turned on, which the import does not touch.

      • namestringrequired
      • profilesarray destringrequired
    • absentarray destringrequired

      The project's services that the model does not have: kept as they are.

    • deploystring | nullrequired

      The deploy the import made (GET /v1/deploys/{id}). Null: a dry run, or nothing would get a version.

      pattern: ^dpl-
  • accountsoptionalarray deobject

    With sole_owner: the accounts the user is the only owner of.

    • idstringrequired
      pattern: ^acct-[0-9a-hjkmnp-tv-z]{16}$
    • namestringrequired

402

account_in_debt: the account owes (its balance is below zero), and until it has paid nothing of it can be used: every call of its keys, and of the tokens of its sandboxes, is refused so, but those that read what it owes and pay it; those that list and read the backups of its volumes and make a volume from one (POST /v1/volumes with from_backup, but while what it stores is being deleted), what one that stopped paying wants most; and, unless the control plane is started with --debt-allows-delete false, those that list, read and delete the volumes, snapshots and sandboxes it stores, and delete its backups. Never a new backup. Before the key's scopes are looked at, and whatever the path. The message says how much is owed, since when, when what the account stores is deleted, how to pay, that it can delete what it stores to stop the amount from growing, and that a volume is still made from a backup (GET /v1/billing, debt).
JSON
  • errorobjectrequired
    • typestringrequired
    • codestringrequired
    • messagestringrequired
    • paramstring | nullrequired
    • retryablebooleanrequired
    • request_idstringrequired
    • docs_urlstring | nullrequired
  • move_idoptionalstring

    With already_moving: the move of the sandbox that is under way.

  • upgrade_idoptionalstring

    With upgrade_in_progress (the key's upgrade that is not over), upgrade_started (the upgrade that has begun) and rollback_gone (the upgrade whose revert is refused): that upgrade.

    pattern: ^upg-[0-9a-hjkmnp-tv-z]{16}$
  • moveoptionalobject

    With move_over, move_going_back and move_taken_over (POST /internal/moves/{id}/cancel): the move, as it stands.

    • objectstringrequired
    • idstringrequired
      pattern: ^mov-[0-9a-hjkmnp-tv-z]{16}$
    • sandbox_idstringrequired
    • project_idstringrequired
    • fromobjectrequired
      • idstringrequired
      • namestring | nullrequired

        Its name; null for a node that is off the books since.

    • toobjectrequired
      • idstringrequired
      • namestring | nullrequired

        Its name; null for a node that is off the books since.

    • epochintegerrequired

      The epoch the table held the sandbox at when the move began: the old node releases it at it, the new one takes it over one on, the old one takes it back two on, and from the object store three on.

      minimum: 1
    • stepstringrequired

      Where it stands: releasing (the old node is asked to release it; the table has it there), taking_over (the new node is asked to take it over; the table has it there), reclaiming (the new node refused it, and the old one is asked to take it back), restoring (the old one kept nothing of it, and takes it over from the object store), done (on the new node), failed (where it was, or back there, or where a failure no time clears or an operator's cancel found it: error says why).

    • whystringrequired

      What made it: the operator's move, an evacuation, a recovery.

    • wakebooleanrequired

      Whether the new node wakes it: as the call said, or as it was (it ran) when the move began.

    • from_storebooleanrequired

      The new node takes it from the object store alone.

    • releaseobject | nullrequired

      What the old node's release answered (sandboxd/openapi/openapi.json, releaseSandbox), without the release's key; null before it answered.

    • takeoverobject | nullrequired

      What the takeover that ended it answered (takeoverSandbox of the node's document): where its disk and memory came from (source), how long each took (ms), memory; null before.

    • errorrequired

      Why it failed, of one that failed (the new node's refusal, where it was taken back to the old node); null, of one that is done; of one under way, what the step it is at was last told where it is asked again.

      Una de estas opciones:

      • object
      • null
    • attemptsintegerrequired

      How many times the step it is at was asked without an answer that settles it.

      minimum: 0
    • next_attempt_atstring | nullrequired

      When that step is asked again; null where it is not waiting.

      format: date-time
    • batchstring | nullrequired

      The evacuation (evc-…) or recovery (rcv-…) it is of.

    • created_atstringrequired
      format: date-time
    • updated_atstringrequired
      format: date-time
    • ended_atstring | nullrequired
      format: date-time
  • lease_untiloptionalstring

    With lease_held: until when the node may still hold the lease it was granted last, and the time after it for the node to have seen it run out (recoverNode is refused until then).

    format: date-time
  • activeoptionalobject | null

    With standby: the active copy, as the standby last read the lease; null while no copy holds it.

    • namestringrequired
    • urlstringrequired
    • termintegerrequired
      minimum: 1
  • allowed_hostsoptionalarray | null

    With secret_host_not_allowed: the hosts the secret may go to, as the secret has them; null where it was given none (the node's SecretHosts).

  • problemsoptionalarray deobject

    With invalid_definition: every problem of the definition, each where it is and what it is.

    • atstringrequired

      Where in the definition, a dotted path (services.web.port); empty for the whole of it.

    • messagestringrequired
  • reportoptionalobject

    With compose_refused: the import's report, which says what is refused, why, and what to change.

    • objectcompose_importrequired
    • namestring | nullrequired

      The model's name: each service's label compose.project. Null: it has none.

    • dry_runbooleanrequired
    • strictbooleanrequired
    • profilesarray destringrequired
    • treatmentsobjectrequired

      How many items of the report have each treatment.

      • doneintegerrequired
        minimum: 0
      • adaptedintegerrequired
        minimum: 0
      • ignoredintegerrequired
        minimum: 0
      • refusedintegerrequired
        minimum: 0
    • itemsarray deobjectrequired

      The model's top-level elements: name, each volume, network, config and secret, extensions.

      • atstringrequired

        Where in the model: of a service, its attribute (ports[0], depends_on.db, environment.KEY), empty for the service as a whole; of the top level, the element (volumes.db-data).

      • treatmentdoneoadaptedoignoredorefusedrequired

        done: it means the same on the platform; adapted: the platform does the same another way; ignored: it matters to a local run alone (refused with strict); refused: the platform cannot do it, and doing without it would let the program go wrong quietly.

      • whystringrequired
      • changestring | nullrequired

        Of one that is refused: what to change. Null otherwise.

    • servicesarray deobjectrequired

      The services the import takes, in the model's order.

      • namestringrequired

        The service's name, which is its app's.

      • app_idstring | nullrequired

        Its app. Null: a dry run of a service the project does not have.

        pattern: ^app-
      • statuscreatedochangedounchangedrequired

        What the import does (or, in a dry run, would do) to its app: makes it, writes what changed, or nothing.

      • kindserviceojobrequired

        job: a service of the model waits for it to complete, and it runs to its end in each deploy that needs it.

      • statefulbooleanrequired

        It mounts a volume of its own: one instance, which has the volume.

      • versionbooleanrequired

        Whether the import's deploy makes a version of it: what its version would be made of, its references resolved, is not what its current version is.

      • itemsarray deobjectrequired
        • atstringrequired

          Where in the model: of a service, its attribute (ports[0], depends_on.db, environment.KEY), empty for the service as a whole; of the top level, the element (volumes.db-data).

        • treatmentdoneoadaptedoignoredorefusedrequired

          done: it means the same on the platform; adapted: the platform does the same another way; ignored: it matters to a local run alone (refused with strict); refused: the platform cannot do it, and doing without it would let the program go wrong quietly.

        • whystringrequired
        • changestring | nullrequired

          Of one that is refused: what to change. Null otherwise.

      • appobjectrequired

        The fields of the service's app as the import writes them: what its versions are made of, and what holds at once.

        • imagestring | nullrequired

          What the app's next version is made of, with the seven fields after it: its newest version's, with what was changed since. null: the app has no image, and no version.

        • image_pull_secretstring | nullrequired

          The name of the secret that holds the credentials of the image's registry. No answer has its value. null: none.

        • start_commandrequired

          Una de estas opciones:

          • StartCommand
          • null
        • start_argsrequired

          What follows the image's ENTRYPOINT in the place of its CMD. null: the image's CMD, or the whole of start_command.

          Una de estas opciones:

          • array<string>
          • null
        • portinteger | nullrequired
        • readyobjectrequired

          How to tell that an instance is up, which a version's template is saved after: exactly one of the three. The first two are asked from outside the instance, so the service has to listen on every address (0.0.0.0).

          • httpoptionalstring

            A path, beginning with /: a GET of it on the app's port is answered with a status from 200 to 399.

          • tcpoptionaltrue

            The app's port takes a connection.

          • commandoptional

            A command that succeeds in the instance. It is run by the instance's shell: a list of arguments is written as one line for it.

            Una de estas opciones:

            • string
            • array<string>
        • envmapa destringrequired
        • secretsmapa destringrequired

          Names of secrets. No answer has a value.

        • sizeobjectrequired

          What one instance may use at most. It is billed by what it uses.

          • vcpusintegerrequired
            minimum: 1
          • memory_bytesintegerrequired

            An even number of MiB (a whole number of the 2 MiB pages a node loads a machine's memory in), in bytes. Any other size is 400 invalid_field (param size.memory_bytes): no node would make an instance of it, as none makes a sandbox of it (the node's SandboxLimitsCreate.memory_bytes).

            minimum: 134217728
        • volumesarray deobjectrequired
          • volumestringrequired
          • pathstringrequired
          • read_onlybooleanrequired
          • nocopyoptionalunknown

            There only where it was given as true.

        • kindstringrequired
        • filesarray deobjectrequired
          • pathstringrequired
          • modeintegerrequired
          • uidintegerrequired
          • gidintegerrequired
          • contentoptionalstring
          • secretoptionalstring
        • userstring | nullrequired
        • working_dirstring | nullrequired
        • stoprequired

          Una de estas opciones:

          • ServiceStop
            • signalstring | nullrequired
            • grace_msinteger | nullrequired
          • null
        • hostnamestring | nullrequired
        • regionsarray destringrequired

          In the order na, eu, ap.

        • scaleobjectrequired
          • min_instancesintegerrequired

            How many instances of the app are kept awake in a region, whether a request comes or not: the floor. The nodes of the region that serve the app share it as they share max_instances (two nodes and a floor of 2: one on each; a floor of 1: one, on the node that is tried first; a floor of 5: three and two), and a node that is away is made up for by the others. A node makes its part when it is told, not at the first request, and neither stops nor deletes those instances for being idle; they are metered as instances that are awake. 0, which is what it is when nothing is said: none. Lowered, the instances beyond it are idle from that moment, and are stopped and deleted as idle instances are. At most max_instances, where that is given, and at most what the account's plan lets an app ask for in a region (caps.app_instances.limit: 5, 50 and 500): more is refused, 403 plan_cap_reached.

            minimum: 0
          • max_instancesinteger | nullrequired

            How many instances the app has at the most in a region: the nodes of the region that serve the app share it. At most what the account's plan lets an app ask for (caps.app_instances.limit: 5 on Free, 500 on Paid): more is refused, 403 plan_cap_reached, and what an app asked for on a plan that let it is kept, and held to the plan's number. null: the plan's number for an app that names none (caps.app_instances.default: 2 on Free, 50 on Paid), or min_instances where that is more. What the nodes are told is no more than the account has room for awake beside everything else of it that is awake, nor less than the app has awake.

            minimum: 1
          • concurrencyinteger | nullrequired

            A cap on what one instance has in hand at a time: the requests the edges sent it, and the connections it has open on its project's network (through its service's address, which the node shares out among the instances connection by connection, or to its own). An instance with that many is full, and another is made for more. null, which is what it is when nothing is said: no cap. A node then adds an instance when the ones it has are near their size in CPU or memory, and an app does not have to guess a number; a service that takes one request at a time gives one.

            maximum: 1000 · minimum: 1
          • sleep_after_msinteger | nullrequired

            The most an instance waits without a request before it stops, in milliseconds. Below it, its node picks the wait by the app's own traffic: the wait that would have cost least over the app's last gaps between a request's end and the next one, a second up and idle against a stop and the resume after it in CPU time, so that an app whose requests come closer together than a stop pays for stays up through them, and one whose requests are far apart stops as soon as its last answer has ended (0: stops at once, always). null, which is what it is when nothing is said: the platform's most, three seconds. An app that said 100, the default before 2026-10-28, says null from then on. One that stays stopped is deleted by its node after delete_after_ms, and the next request that needs an instance gets a new one; the instances min_instances keeps awake are neither stopped nor deleted.

            maximum: 600000 · minimum: 0
          • sleep_notice_msinteger | nullrequired

            How long before an instance stops for being idle the programs connected to its guest's /run/sbx/awake are sent the line sleep N, in milliseconds: they flush what they hold, close their connections, or raise a hold, which puts the stop off. The stop comes that long after, whether they answer or not; where nobody is connected, it is not put off. A keyed instance is told as well. 0 or null: not told (the default).

            maximum: 60000 · minimum: 0
          • sleep_hold_max_msintegerrequired

            How long the holds a program in an instance raises keep it up after its time idle has run out, in milliseconds: a program connects to /run/sbx/awake and sends + before it answers a request whose work goes on after the answer, and - (or closes the connection, or dies) when that work is done; the instance stops sleep_after_ms after the last hold is lowered. Past this time it stops whatever holds it, the programs connected told sleep N cap first where sleep_notice_ms asks for a notice. Held, it runs and is metered as running. 0: holds are not heeded. Null: 30000 (the default).

            maximum: 900000 · minimum: 0
          • delete_after_msinteger | nullrequired

            After how long idle an instance is deleted, counted from where its stop is counted from: the next request that needs one gets a new instance, a clone of the version's template, without what the deleted one had gathered in its memory and on its disk (keep what is to last in a volume or a database). A second to an hour, the platform's time, which is the most: more is refused (400 invalid_field), and an instance that is to stay warm is one of min_instances. null, which is what it is when nothing is said: the hour of the nodes (--app-delete-after-seconds). A change holds for the instances the app has as for the ones made after: each node judges every instance by the app's time as it is, once a second at the least. Not for the app's keyed instances, which keys.delete_after_idle_ms is for.

            maximum: 3600000 · minimum: 1000
          • suspend_after_msoptionalinteger | null

            How long a stopped instance stays paused before its memory is saved and it sleeps for real, in milliseconds: paused, the next request is answered in a millisecond or two and its memory is metered; asleep, it wakes in a few more and only what it keeps is. Its node holds it to the longest it holds any instance paused (ten minutes); the host's memory running short, a slot another machine needs and a sleep somebody asks for put it to sleep sooner. Null (the default): its node's longest. A database's suspend_after_seconds is its app's.

            maximum: 86400000 · minimum: 1000
          • wake_daily_atoptionalstring | null

            A time of day in UTC, HH:MM, at which the app's keyed instances and its stateful service's instance that rest are woken every day, for work of their own that waits for a wake (a database's maintenance, say); the instances the nodes make and delete by the app's requests are woken by requests alone. The wake is no activity: what runs then is, and once that is done the instance rests again. A stateful service's instance is woken besides for the next run of its PostgreSQL server's pg_cron jobs, whatever this says. Null (the default): none. A database's wake_daily_at is its app's.

            pattern: ^([01][0-9]|2[0-3]):[0-5][0-9]$
        • keysrequired

          Una de estas opciones:

          • object
          • null
        • egressobject | nullrequired

          Egress rules as a sandbox has them (default, allow, deny, http, inject), null when there are none; kept as the calls gave them (a PATCH merges them key by key, as a node does). They hold at once and make no version: the nodes that serve the app put them on every instance of its group, the ones there are (one that runs for the connections it has too) and the ones made after; a keyed instance does not have them. The control plane checks only that they are an object of those keys of at most 256 KiB: what the rules say is a node's to check, when it applies them, and an instance whose rules its node cannot read is refused everything (the node's log says why).

          • defaultoptionalallowodenyonull
          • allowoptionalarray | null
          • denyoptionalarray | null
          • httpoptionalarray | null
          • injectoptionalarray | null
          • deny_portsoptionalarray | null

            Ports no connection reaches, TCP or UDP, wherever it goes and whatever else the rules say. A Browsers session's always hold 25, 465, 587 and the ports mining pools listen on (3333, 4444, 5555, 7777, 14444, 45700): the control plane adds them to whatever a change of its rules gives.

        • metadatamapa destringrequired
        • logsobjectrequired

          What the app's log has, each part as it was given or at its default (AppLogSettings).

          • outputbooleanrequired
          • keepbooleanrequired
          • requestsobjectrequired
            • sample_ratenumberrequired
              maximum: 1 · minimum: 0
            • keep_errorsbooleanrequired
        • publicbooleanrequired

          Whether it answers at its default domain (url).

        • depends_onarray deobjectrequired
          • servicestringrequired
          • conditionstringrequired
        • networkobjectrequired
          • aliasesarray destringrequired
          • endpoint_modestringrequired
        • drivesoptionalarray deobject

          The Drives its instance mounts (AppCreate.drives): none, [], where it names none.

          • drivestringrequired
          • pathstringrequired
          • modestringrequired
        • key_drivesoptionalarray deobject

          The Drives each key's instance has of its own (AppCreate.key_drives): none, [], where it gives none.

          • namestringrequired
          • pathstringrequired
          • size_bytesinteger | nullrequired
            format: int64
          • nocopyoptionalunknown

            There only where it was given as true.

        • runtimeoptional

          Its runtime settings (AppCreate.runtime). null: none.

          Una de estas opciones:

          • Runtime
            • ulimitsmapa deRuntimeLimitGivenrequired
            • sysctlsmapa destringrequired
            • tmpfsarray deobjectrequired
              • pathstringrequired
              • size_bytesinteger | nullrequired
              • modeintegerrequired
            • dnsobjectrequired
              • serversarray destringrequired
              • searcharray destringrequired
              • optionsarray destringrequired
            • shm_size_bytesinteger | nullrequired
          • null
        • tracesoptionalobject

          What the app's requests' traces have, each part as it was given or at its default (AppTraceSettings): no span where sample_rate is 0.

          • sample_ratenumberrequired
            maximum: 1 · minimum: 0
          • trust_incomingbooleanrequired
    • volumesarray deobjectrequired

      The volumes of the project the services mount.

      • modelstring | nullrequired

        Its name in the model. Null: made for a path a stateful service mounts with no volume named.

      • namestringrequired

        Its name in the project.

      • idstring | nullrequired

        Null: not made yet (to_make).

        pattern: ^vol-
      • sharedbooleanrequired

        Several services mount it, or one with several instances: every instance of each mounts it at once.

      • statusto_makeomadeothereoexternalrequired

        to_make: the import makes it (a dry run says so of what it would make); made: this import made it; there: an earlier import made it, or a volume of its name was there; external: the model names it external, and the project has it.

      • used_byarray destringrequired
    • secretsarray deobjectrequired

      The secrets of the model that services are granted.

      • modelstringrequired

        Its name in the model.

      • namestringrequired

        The project's secret that holds it: the model's name, with each character a secret's name does not take written as an underscore.

      • therebooleanrequired

        Whether the project has it. An import (not a dry run) with one that is not there is refused 404 secret_not_found.

      • used_byarray destringrequired
    • left_outarray deobjectrequired

      The services of profiles that are not turned on, which the import does not touch.

      • namestringrequired
      • profilesarray destringrequired
    • absentarray destringrequired

      The project's services that the model does not have: kept as they are.

    • deploystring | nullrequired

      The deploy the import made (GET /v1/deploys/{id}). Null: a dry run, or nothing would get a version.

      pattern: ^dpl-
  • accountsoptionalarray deobject

    With sole_owner: the accounts the user is the only owner of.

    • idstringrequired
      pattern: ^acct-[0-9a-hjkmnp-tv-z]{16}$
    • namestringrequired

403

key_scope: the key does not allow write, or the call gives the app a secret it does not have and the key does not allow admin; needs_api_key; app_limit_reached: the account has as many apps as its plan lets it have; env_limit_reached: more environment variables and secrets together than the plan lets an app have; secret_limit_reached: the version the call would make gives its instances more of the project's secrets, as environment variables and files together, than the plan lets one sandbox be given (max_sandbox_secrets_bytes of GET /v1/billing), and more than they are given now (param is the place at which they pass it: secrets.<VARIABLE> or files[n].secret; of a project's deploy, services.<name>. before it); the message names the limit; plan_size_exceeded: an instance larger than the plan's largest sandbox (param is size); plan_cap_reached: scale.max_instances or scale.min_instances given as more than the plan lets an app ask for in a region (param is app_instances; the message has the plan's number), or as many apps and versions made in the last hour as the account may make (param is app_creates_per_hour; retryable, with Retry-After); spend_limit_reached, balance_used_up: a call that would make a version (a create with an image, a change of what a version is made of, a deploy) in a project that is held back, refused with the code and the words a create of a sandbox gets there, before anything is built; one that makes no version is taken, and so is the same call once the project is let go of; account_held: the account is held for its payments, and every call of its keys but GET /v1/account is refused so, before the scopes are looked at (x-error-codes); secret_host_not_allowed: the app's egress rules as the call would leave them send a secret by inject to a host it may not go to (param is egress.inject[n].host; looked at where the call gives rules or makes a version), or the version it would make pulls its image with a secret that may not go to the image's registry (param is image_pull_secret); beside error, allowed_hosts says where the secret may go.
JSON
  • errorobjectrequired
    • typestringrequired
    • codestringrequired
    • messagestringrequired
    • paramstring | nullrequired
    • retryablebooleanrequired
    • request_idstringrequired
    • docs_urlstring | nullrequired
  • move_idoptionalstring

    With already_moving: the move of the sandbox that is under way.

  • upgrade_idoptionalstring

    With upgrade_in_progress (the key's upgrade that is not over), upgrade_started (the upgrade that has begun) and rollback_gone (the upgrade whose revert is refused): that upgrade.

    pattern: ^upg-[0-9a-hjkmnp-tv-z]{16}$
  • moveoptionalobject

    With move_over, move_going_back and move_taken_over (POST /internal/moves/{id}/cancel): the move, as it stands.

    • objectstringrequired
    • idstringrequired
      pattern: ^mov-[0-9a-hjkmnp-tv-z]{16}$
    • sandbox_idstringrequired
    • project_idstringrequired
    • fromobjectrequired
      • idstringrequired
      • namestring | nullrequired

        Its name; null for a node that is off the books since.

    • toobjectrequired
      • idstringrequired
      • namestring | nullrequired

        Its name; null for a node that is off the books since.

    • epochintegerrequired

      The epoch the table held the sandbox at when the move began: the old node releases it at it, the new one takes it over one on, the old one takes it back two on, and from the object store three on.

      minimum: 1
    • stepstringrequired

      Where it stands: releasing (the old node is asked to release it; the table has it there), taking_over (the new node is asked to take it over; the table has it there), reclaiming (the new node refused it, and the old one is asked to take it back), restoring (the old one kept nothing of it, and takes it over from the object store), done (on the new node), failed (where it was, or back there, or where a failure no time clears or an operator's cancel found it: error says why).

    • whystringrequired

      What made it: the operator's move, an evacuation, a recovery.

    • wakebooleanrequired

      Whether the new node wakes it: as the call said, or as it was (it ran) when the move began.

    • from_storebooleanrequired

      The new node takes it from the object store alone.

    • releaseobject | nullrequired

      What the old node's release answered (sandboxd/openapi/openapi.json, releaseSandbox), without the release's key; null before it answered.

    • takeoverobject | nullrequired

      What the takeover that ended it answered (takeoverSandbox of the node's document): where its disk and memory came from (source), how long each took (ms), memory; null before.

    • errorrequired

      Why it failed, of one that failed (the new node's refusal, where it was taken back to the old node); null, of one that is done; of one under way, what the step it is at was last told where it is asked again.

      Una de estas opciones:

      • object
      • null
    • attemptsintegerrequired

      How many times the step it is at was asked without an answer that settles it.

      minimum: 0
    • next_attempt_atstring | nullrequired

      When that step is asked again; null where it is not waiting.

      format: date-time
    • batchstring | nullrequired

      The evacuation (evc-…) or recovery (rcv-…) it is of.

    • created_atstringrequired
      format: date-time
    • updated_atstringrequired
      format: date-time
    • ended_atstring | nullrequired
      format: date-time
  • lease_untiloptionalstring

    With lease_held: until when the node may still hold the lease it was granted last, and the time after it for the node to have seen it run out (recoverNode is refused until then).

    format: date-time
  • activeoptionalobject | null

    With standby: the active copy, as the standby last read the lease; null while no copy holds it.

    • namestringrequired
    • urlstringrequired
    • termintegerrequired
      minimum: 1
  • allowed_hostsoptionalarray | null

    With secret_host_not_allowed: the hosts the secret may go to, as the secret has them; null where it was given none (the node's SecretHosts).

  • problemsoptionalarray deobject

    With invalid_definition: every problem of the definition, each where it is and what it is.

    • atstringrequired

      Where in the definition, a dotted path (services.web.port); empty for the whole of it.

    • messagestringrequired
  • reportoptionalobject

    With compose_refused: the import's report, which says what is refused, why, and what to change.

    • objectcompose_importrequired
    • namestring | nullrequired

      The model's name: each service's label compose.project. Null: it has none.

    • dry_runbooleanrequired
    • strictbooleanrequired
    • profilesarray destringrequired
    • treatmentsobjectrequired

      How many items of the report have each treatment.

      • doneintegerrequired
        minimum: 0
      • adaptedintegerrequired
        minimum: 0
      • ignoredintegerrequired
        minimum: 0
      • refusedintegerrequired
        minimum: 0
    • itemsarray deobjectrequired

      The model's top-level elements: name, each volume, network, config and secret, extensions.

      • atstringrequired

        Where in the model: of a service, its attribute (ports[0], depends_on.db, environment.KEY), empty for the service as a whole; of the top level, the element (volumes.db-data).

      • treatmentdoneoadaptedoignoredorefusedrequired

        done: it means the same on the platform; adapted: the platform does the same another way; ignored: it matters to a local run alone (refused with strict); refused: the platform cannot do it, and doing without it would let the program go wrong quietly.

      • whystringrequired
      • changestring | nullrequired

        Of one that is refused: what to change. Null otherwise.

    • servicesarray deobjectrequired

      The services the import takes, in the model's order.

      • namestringrequired

        The service's name, which is its app's.

      • app_idstring | nullrequired

        Its app. Null: a dry run of a service the project does not have.

        pattern: ^app-
      • statuscreatedochangedounchangedrequired

        What the import does (or, in a dry run, would do) to its app: makes it, writes what changed, or nothing.

      • kindserviceojobrequired

        job: a service of the model waits for it to complete, and it runs to its end in each deploy that needs it.

      • statefulbooleanrequired

        It mounts a volume of its own: one instance, which has the volume.

      • versionbooleanrequired

        Whether the import's deploy makes a version of it: what its version would be made of, its references resolved, is not what its current version is.

      • itemsarray deobjectrequired
        • atstringrequired

          Where in the model: of a service, its attribute (ports[0], depends_on.db, environment.KEY), empty for the service as a whole; of the top level, the element (volumes.db-data).

        • treatmentdoneoadaptedoignoredorefusedrequired

          done: it means the same on the platform; adapted: the platform does the same another way; ignored: it matters to a local run alone (refused with strict); refused: the platform cannot do it, and doing without it would let the program go wrong quietly.

        • whystringrequired
        • changestring | nullrequired

          Of one that is refused: what to change. Null otherwise.

      • appobjectrequired

        The fields of the service's app as the import writes them: what its versions are made of, and what holds at once.

        • imagestring | nullrequired

          What the app's next version is made of, with the seven fields after it: its newest version's, with what was changed since. null: the app has no image, and no version.

        • image_pull_secretstring | nullrequired

          The name of the secret that holds the credentials of the image's registry. No answer has its value. null: none.

        • start_commandrequired

          Una de estas opciones:

          • StartCommand
          • null
        • start_argsrequired

          What follows the image's ENTRYPOINT in the place of its CMD. null: the image's CMD, or the whole of start_command.

          Una de estas opciones:

          • array<string>
          • null
        • portinteger | nullrequired
        • readyobjectrequired

          How to tell that an instance is up, which a version's template is saved after: exactly one of the three. The first two are asked from outside the instance, so the service has to listen on every address (0.0.0.0).

          • httpoptionalstring

            A path, beginning with /: a GET of it on the app's port is answered with a status from 200 to 399.

          • tcpoptionaltrue

            The app's port takes a connection.

          • commandoptional

            A command that succeeds in the instance. It is run by the instance's shell: a list of arguments is written as one line for it.

            Una de estas opciones:

            • string
            • array<string>
        • envmapa destringrequired
        • secretsmapa destringrequired

          Names of secrets. No answer has a value.

        • sizeobjectrequired

          What one instance may use at most. It is billed by what it uses.

          • vcpusintegerrequired
            minimum: 1
          • memory_bytesintegerrequired

            An even number of MiB (a whole number of the 2 MiB pages a node loads a machine's memory in), in bytes. Any other size is 400 invalid_field (param size.memory_bytes): no node would make an instance of it, as none makes a sandbox of it (the node's SandboxLimitsCreate.memory_bytes).

            minimum: 134217728
        • volumesarray deobjectrequired
          • volumestringrequired
          • pathstringrequired
          • read_onlybooleanrequired
          • nocopyoptionalunknown

            There only where it was given as true.

        • kindstringrequired
        • filesarray deobjectrequired
          • pathstringrequired
          • modeintegerrequired
          • uidintegerrequired
          • gidintegerrequired
          • contentoptionalstring
          • secretoptionalstring
        • userstring | nullrequired
        • working_dirstring | nullrequired
        • stoprequired

          Una de estas opciones:

          • ServiceStop
            • signalstring | nullrequired
            • grace_msinteger | nullrequired
          • null
        • hostnamestring | nullrequired
        • regionsarray destringrequired

          In the order na, eu, ap.

        • scaleobjectrequired
          • min_instancesintegerrequired

            How many instances of the app are kept awake in a region, whether a request comes or not: the floor. The nodes of the region that serve the app share it as they share max_instances (two nodes and a floor of 2: one on each; a floor of 1: one, on the node that is tried first; a floor of 5: three and two), and a node that is away is made up for by the others. A node makes its part when it is told, not at the first request, and neither stops nor deletes those instances for being idle; they are metered as instances that are awake. 0, which is what it is when nothing is said: none. Lowered, the instances beyond it are idle from that moment, and are stopped and deleted as idle instances are. At most max_instances, where that is given, and at most what the account's plan lets an app ask for in a region (caps.app_instances.limit: 5, 50 and 500): more is refused, 403 plan_cap_reached.

            minimum: 0
          • max_instancesinteger | nullrequired

            How many instances the app has at the most in a region: the nodes of the region that serve the app share it. At most what the account's plan lets an app ask for (caps.app_instances.limit: 5 on Free, 500 on Paid): more is refused, 403 plan_cap_reached, and what an app asked for on a plan that let it is kept, and held to the plan's number. null: the plan's number for an app that names none (caps.app_instances.default: 2 on Free, 50 on Paid), or min_instances where that is more. What the nodes are told is no more than the account has room for awake beside everything else of it that is awake, nor less than the app has awake.

            minimum: 1
          • concurrencyinteger | nullrequired

            A cap on what one instance has in hand at a time: the requests the edges sent it, and the connections it has open on its project's network (through its service's address, which the node shares out among the instances connection by connection, or to its own). An instance with that many is full, and another is made for more. null, which is what it is when nothing is said: no cap. A node then adds an instance when the ones it has are near their size in CPU or memory, and an app does not have to guess a number; a service that takes one request at a time gives one.

            maximum: 1000 · minimum: 1
          • sleep_after_msinteger | nullrequired

            The most an instance waits without a request before it stops, in milliseconds. Below it, its node picks the wait by the app's own traffic: the wait that would have cost least over the app's last gaps between a request's end and the next one, a second up and idle against a stop and the resume after it in CPU time, so that an app whose requests come closer together than a stop pays for stays up through them, and one whose requests are far apart stops as soon as its last answer has ended (0: stops at once, always). null, which is what it is when nothing is said: the platform's most, three seconds. An app that said 100, the default before 2026-10-28, says null from then on. One that stays stopped is deleted by its node after delete_after_ms, and the next request that needs an instance gets a new one; the instances min_instances keeps awake are neither stopped nor deleted.

            maximum: 600000 · minimum: 0
          • sleep_notice_msinteger | nullrequired

            How long before an instance stops for being idle the programs connected to its guest's /run/sbx/awake are sent the line sleep N, in milliseconds: they flush what they hold, close their connections, or raise a hold, which puts the stop off. The stop comes that long after, whether they answer or not; where nobody is connected, it is not put off. A keyed instance is told as well. 0 or null: not told (the default).

            maximum: 60000 · minimum: 0
          • sleep_hold_max_msintegerrequired

            How long the holds a program in an instance raises keep it up after its time idle has run out, in milliseconds: a program connects to /run/sbx/awake and sends + before it answers a request whose work goes on after the answer, and - (or closes the connection, or dies) when that work is done; the instance stops sleep_after_ms after the last hold is lowered. Past this time it stops whatever holds it, the programs connected told sleep N cap first where sleep_notice_ms asks for a notice. Held, it runs and is metered as running. 0: holds are not heeded. Null: 30000 (the default).

            maximum: 900000 · minimum: 0
          • delete_after_msinteger | nullrequired

            After how long idle an instance is deleted, counted from where its stop is counted from: the next request that needs one gets a new instance, a clone of the version's template, without what the deleted one had gathered in its memory and on its disk (keep what is to last in a volume or a database). A second to an hour, the platform's time, which is the most: more is refused (400 invalid_field), and an instance that is to stay warm is one of min_instances. null, which is what it is when nothing is said: the hour of the nodes (--app-delete-after-seconds). A change holds for the instances the app has as for the ones made after: each node judges every instance by the app's time as it is, once a second at the least. Not for the app's keyed instances, which keys.delete_after_idle_ms is for.

            maximum: 3600000 · minimum: 1000
          • suspend_after_msoptionalinteger | null

            How long a stopped instance stays paused before its memory is saved and it sleeps for real, in milliseconds: paused, the next request is answered in a millisecond or two and its memory is metered; asleep, it wakes in a few more and only what it keeps is. Its node holds it to the longest it holds any instance paused (ten minutes); the host's memory running short, a slot another machine needs and a sleep somebody asks for put it to sleep sooner. Null (the default): its node's longest. A database's suspend_after_seconds is its app's.

            maximum: 86400000 · minimum: 1000
          • wake_daily_atoptionalstring | null

            A time of day in UTC, HH:MM, at which the app's keyed instances and its stateful service's instance that rest are woken every day, for work of their own that waits for a wake (a database's maintenance, say); the instances the nodes make and delete by the app's requests are woken by requests alone. The wake is no activity: what runs then is, and once that is done the instance rests again. A stateful service's instance is woken besides for the next run of its PostgreSQL server's pg_cron jobs, whatever this says. Null (the default): none. A database's wake_daily_at is its app's.

            pattern: ^([01][0-9]|2[0-3]):[0-5][0-9]$
        • keysrequired

          Una de estas opciones:

          • object
          • null
        • egressobject | nullrequired

          Egress rules as a sandbox has them (default, allow, deny, http, inject), null when there are none; kept as the calls gave them (a PATCH merges them key by key, as a node does). They hold at once and make no version: the nodes that serve the app put them on every instance of its group, the ones there are (one that runs for the connections it has too) and the ones made after; a keyed instance does not have them. The control plane checks only that they are an object of those keys of at most 256 KiB: what the rules say is a node's to check, when it applies them, and an instance whose rules its node cannot read is refused everything (the node's log says why).

          • defaultoptionalallowodenyonull
          • allowoptionalarray | null
          • denyoptionalarray | null
          • httpoptionalarray | null
          • injectoptionalarray | null
          • deny_portsoptionalarray | null

            Ports no connection reaches, TCP or UDP, wherever it goes and whatever else the rules say. A Browsers session's always hold 25, 465, 587 and the ports mining pools listen on (3333, 4444, 5555, 7777, 14444, 45700): the control plane adds them to whatever a change of its rules gives.

        • metadatamapa destringrequired
        • logsobjectrequired

          What the app's log has, each part as it was given or at its default (AppLogSettings).

          • outputbooleanrequired
          • keepbooleanrequired
          • requestsobjectrequired
            • sample_ratenumberrequired
              maximum: 1 · minimum: 0
            • keep_errorsbooleanrequired
        • publicbooleanrequired

          Whether it answers at its default domain (url).

        • depends_onarray deobjectrequired
          • servicestringrequired
          • conditionstringrequired
        • networkobjectrequired
          • aliasesarray destringrequired
          • endpoint_modestringrequired
        • drivesoptionalarray deobject

          The Drives its instance mounts (AppCreate.drives): none, [], where it names none.

          • drivestringrequired
          • pathstringrequired
          • modestringrequired
        • key_drivesoptionalarray deobject

          The Drives each key's instance has of its own (AppCreate.key_drives): none, [], where it gives none.

          • namestringrequired
          • pathstringrequired
          • size_bytesinteger | nullrequired
            format: int64
          • nocopyoptionalunknown

            There only where it was given as true.

        • runtimeoptional

          Its runtime settings (AppCreate.runtime). null: none.

          Una de estas opciones:

          • Runtime
            • ulimitsmapa deRuntimeLimitGivenrequired
            • sysctlsmapa destringrequired
            • tmpfsarray deobjectrequired
              • pathstringrequired
              • size_bytesinteger | nullrequired
              • modeintegerrequired
            • dnsobjectrequired
              • serversarray destringrequired
              • searcharray destringrequired
              • optionsarray destringrequired
            • shm_size_bytesinteger | nullrequired
          • null
        • tracesoptionalobject

          What the app's requests' traces have, each part as it was given or at its default (AppTraceSettings): no span where sample_rate is 0.

          • sample_ratenumberrequired
            maximum: 1 · minimum: 0
          • trust_incomingbooleanrequired
    • volumesarray deobjectrequired

      The volumes of the project the services mount.

      • modelstring | nullrequired

        Its name in the model. Null: made for a path a stateful service mounts with no volume named.

      • namestringrequired

        Its name in the project.

      • idstring | nullrequired

        Null: not made yet (to_make).

        pattern: ^vol-
      • sharedbooleanrequired

        Several services mount it, or one with several instances: every instance of each mounts it at once.

      • statusto_makeomadeothereoexternalrequired

        to_make: the import makes it (a dry run says so of what it would make); made: this import made it; there: an earlier import made it, or a volume of its name was there; external: the model names it external, and the project has it.

      • used_byarray destringrequired
    • secretsarray deobjectrequired

      The secrets of the model that services are granted.

      • modelstringrequired

        Its name in the model.

      • namestringrequired

        The project's secret that holds it: the model's name, with each character a secret's name does not take written as an underscore.

      • therebooleanrequired

        Whether the project has it. An import (not a dry run) with one that is not there is refused 404 secret_not_found.

      • used_byarray destringrequired
    • left_outarray deobjectrequired

      The services of profiles that are not turned on, which the import does not touch.

      • namestringrequired
      • profilesarray destringrequired
    • absentarray destringrequired

      The project's services that the model does not have: kept as they are.

    • deploystring | nullrequired

      The deploy the import made (GET /v1/deploys/{id}). Null: a dry run, or nothing would get a version.

      pattern: ^dpl-
  • accountsoptionalarray deobject

    With sole_owner: the accounts the user is the only owner of.

    • idstringrequired
      pattern: ^acct-[0-9a-hjkmnp-tv-z]{16}$
    • namestringrequired

404

app_not_found; secret_not_found; volume_not_found. drive_not_found: drives names a Drive the project does not have, or a volume of another kind (param drives[i].drive).
JSON
  • errorobjectrequired
    • typestringrequired
    • codestringrequired
    • messagestringrequired
    • paramstring | nullrequired
    • retryablebooleanrequired
    • request_idstringrequired
    • docs_urlstring | nullrequired
  • move_idoptionalstring

    With already_moving: the move of the sandbox that is under way.

  • upgrade_idoptionalstring

    With upgrade_in_progress (the key's upgrade that is not over), upgrade_started (the upgrade that has begun) and rollback_gone (the upgrade whose revert is refused): that upgrade.

    pattern: ^upg-[0-9a-hjkmnp-tv-z]{16}$
  • moveoptionalobject

    With move_over, move_going_back and move_taken_over (POST /internal/moves/{id}/cancel): the move, as it stands.

    • objectstringrequired
    • idstringrequired
      pattern: ^mov-[0-9a-hjkmnp-tv-z]{16}$
    • sandbox_idstringrequired
    • project_idstringrequired
    • fromobjectrequired
      • idstringrequired
      • namestring | nullrequired

        Its name; null for a node that is off the books since.

    • toobjectrequired
      • idstringrequired
      • namestring | nullrequired

        Its name; null for a node that is off the books since.

    • epochintegerrequired

      The epoch the table held the sandbox at when the move began: the old node releases it at it, the new one takes it over one on, the old one takes it back two on, and from the object store three on.

      minimum: 1
    • stepstringrequired

      Where it stands: releasing (the old node is asked to release it; the table has it there), taking_over (the new node is asked to take it over; the table has it there), reclaiming (the new node refused it, and the old one is asked to take it back), restoring (the old one kept nothing of it, and takes it over from the object store), done (on the new node), failed (where it was, or back there, or where a failure no time clears or an operator's cancel found it: error says why).

    • whystringrequired

      What made it: the operator's move, an evacuation, a recovery.

    • wakebooleanrequired

      Whether the new node wakes it: as the call said, or as it was (it ran) when the move began.

    • from_storebooleanrequired

      The new node takes it from the object store alone.

    • releaseobject | nullrequired

      What the old node's release answered (sandboxd/openapi/openapi.json, releaseSandbox), without the release's key; null before it answered.

    • takeoverobject | nullrequired

      What the takeover that ended it answered (takeoverSandbox of the node's document): where its disk and memory came from (source), how long each took (ms), memory; null before.

    • errorrequired

      Why it failed, of one that failed (the new node's refusal, where it was taken back to the old node); null, of one that is done; of one under way, what the step it is at was last told where it is asked again.

      Una de estas opciones:

      • object
      • null
    • attemptsintegerrequired

      How many times the step it is at was asked without an answer that settles it.

      minimum: 0
    • next_attempt_atstring | nullrequired

      When that step is asked again; null where it is not waiting.

      format: date-time
    • batchstring | nullrequired

      The evacuation (evc-…) or recovery (rcv-…) it is of.

    • created_atstringrequired
      format: date-time
    • updated_atstringrequired
      format: date-time
    • ended_atstring | nullrequired
      format: date-time
  • lease_untiloptionalstring

    With lease_held: until when the node may still hold the lease it was granted last, and the time after it for the node to have seen it run out (recoverNode is refused until then).

    format: date-time
  • activeoptionalobject | null

    With standby: the active copy, as the standby last read the lease; null while no copy holds it.

    • namestringrequired
    • urlstringrequired
    • termintegerrequired
      minimum: 1
  • allowed_hostsoptionalarray | null

    With secret_host_not_allowed: the hosts the secret may go to, as the secret has them; null where it was given none (the node's SecretHosts).

  • problemsoptionalarray deobject

    With invalid_definition: every problem of the definition, each where it is and what it is.

    • atstringrequired

      Where in the definition, a dotted path (services.web.port); empty for the whole of it.

    • messagestringrequired
  • reportoptionalobject

    With compose_refused: the import's report, which says what is refused, why, and what to change.

    • objectcompose_importrequired
    • namestring | nullrequired

      The model's name: each service's label compose.project. Null: it has none.

    • dry_runbooleanrequired
    • strictbooleanrequired
    • profilesarray destringrequired
    • treatmentsobjectrequired

      How many items of the report have each treatment.

      • doneintegerrequired
        minimum: 0
      • adaptedintegerrequired
        minimum: 0
      • ignoredintegerrequired
        minimum: 0
      • refusedintegerrequired
        minimum: 0
    • itemsarray deobjectrequired

      The model's top-level elements: name, each volume, network, config and secret, extensions.

      • atstringrequired

        Where in the model: of a service, its attribute (ports[0], depends_on.db, environment.KEY), empty for the service as a whole; of the top level, the element (volumes.db-data).

      • treatmentdoneoadaptedoignoredorefusedrequired

        done: it means the same on the platform; adapted: the platform does the same another way; ignored: it matters to a local run alone (refused with strict); refused: the platform cannot do it, and doing without it would let the program go wrong quietly.

      • whystringrequired
      • changestring | nullrequired

        Of one that is refused: what to change. Null otherwise.

    • servicesarray deobjectrequired

      The services the import takes, in the model's order.

      • namestringrequired

        The service's name, which is its app's.

      • app_idstring | nullrequired

        Its app. Null: a dry run of a service the project does not have.

        pattern: ^app-
      • statuscreatedochangedounchangedrequired

        What the import does (or, in a dry run, would do) to its app: makes it, writes what changed, or nothing.

      • kindserviceojobrequired

        job: a service of the model waits for it to complete, and it runs to its end in each deploy that needs it.

      • statefulbooleanrequired

        It mounts a volume of its own: one instance, which has the volume.

      • versionbooleanrequired

        Whether the import's deploy makes a version of it: what its version would be made of, its references resolved, is not what its current version is.

      • itemsarray deobjectrequired
        • atstringrequired

          Where in the model: of a service, its attribute (ports[0], depends_on.db, environment.KEY), empty for the service as a whole; of the top level, the element (volumes.db-data).

        • treatmentdoneoadaptedoignoredorefusedrequired

          done: it means the same on the platform; adapted: the platform does the same another way; ignored: it matters to a local run alone (refused with strict); refused: the platform cannot do it, and doing without it would let the program go wrong quietly.

        • whystringrequired
        • changestring | nullrequired

          Of one that is refused: what to change. Null otherwise.

      • appobjectrequired

        The fields of the service's app as the import writes them: what its versions are made of, and what holds at once.

        • imagestring | nullrequired

          What the app's next version is made of, with the seven fields after it: its newest version's, with what was changed since. null: the app has no image, and no version.

        • image_pull_secretstring | nullrequired

          The name of the secret that holds the credentials of the image's registry. No answer has its value. null: none.

        • start_commandrequired

          Una de estas opciones:

          • StartCommand
          • null
        • start_argsrequired

          What follows the image's ENTRYPOINT in the place of its CMD. null: the image's CMD, or the whole of start_command.

          Una de estas opciones:

          • array<string>
          • null
        • portinteger | nullrequired
        • readyobjectrequired

          How to tell that an instance is up, which a version's template is saved after: exactly one of the three. The first two are asked from outside the instance, so the service has to listen on every address (0.0.0.0).

          • httpoptionalstring

            A path, beginning with /: a GET of it on the app's port is answered with a status from 200 to 399.

          • tcpoptionaltrue

            The app's port takes a connection.

          • commandoptional

            A command that succeeds in the instance. It is run by the instance's shell: a list of arguments is written as one line for it.

            Una de estas opciones:

            • string
            • array<string>
        • envmapa destringrequired
        • secretsmapa destringrequired

          Names of secrets. No answer has a value.

        • sizeobjectrequired

          What one instance may use at most. It is billed by what it uses.

          • vcpusintegerrequired
            minimum: 1
          • memory_bytesintegerrequired

            An even number of MiB (a whole number of the 2 MiB pages a node loads a machine's memory in), in bytes. Any other size is 400 invalid_field (param size.memory_bytes): no node would make an instance of it, as none makes a sandbox of it (the node's SandboxLimitsCreate.memory_bytes).

            minimum: 134217728
        • volumesarray deobjectrequired
          • volumestringrequired
          • pathstringrequired
          • read_onlybooleanrequired
          • nocopyoptionalunknown

            There only where it was given as true.

        • kindstringrequired
        • filesarray deobjectrequired
          • pathstringrequired
          • modeintegerrequired
          • uidintegerrequired
          • gidintegerrequired
          • contentoptionalstring
          • secretoptionalstring
        • userstring | nullrequired
        • working_dirstring | nullrequired
        • stoprequired

          Una de estas opciones:

          • ServiceStop
            • signalstring | nullrequired
            • grace_msinteger | nullrequired
          • null
        • hostnamestring | nullrequired
        • regionsarray destringrequired

          In the order na, eu, ap.

        • scaleobjectrequired
          • min_instancesintegerrequired

            How many instances of the app are kept awake in a region, whether a request comes or not: the floor. The nodes of the region that serve the app share it as they share max_instances (two nodes and a floor of 2: one on each; a floor of 1: one, on the node that is tried first; a floor of 5: three and two), and a node that is away is made up for by the others. A node makes its part when it is told, not at the first request, and neither stops nor deletes those instances for being idle; they are metered as instances that are awake. 0, which is what it is when nothing is said: none. Lowered, the instances beyond it are idle from that moment, and are stopped and deleted as idle instances are. At most max_instances, where that is given, and at most what the account's plan lets an app ask for in a region (caps.app_instances.limit: 5, 50 and 500): more is refused, 403 plan_cap_reached.

            minimum: 0
          • max_instancesinteger | nullrequired

            How many instances the app has at the most in a region: the nodes of the region that serve the app share it. At most what the account's plan lets an app ask for (caps.app_instances.limit: 5 on Free, 500 on Paid): more is refused, 403 plan_cap_reached, and what an app asked for on a plan that let it is kept, and held to the plan's number. null: the plan's number for an app that names none (caps.app_instances.default: 2 on Free, 50 on Paid), or min_instances where that is more. What the nodes are told is no more than the account has room for awake beside everything else of it that is awake, nor less than the app has awake.

            minimum: 1
          • concurrencyinteger | nullrequired

            A cap on what one instance has in hand at a time: the requests the edges sent it, and the connections it has open on its project's network (through its service's address, which the node shares out among the instances connection by connection, or to its own). An instance with that many is full, and another is made for more. null, which is what it is when nothing is said: no cap. A node then adds an instance when the ones it has are near their size in CPU or memory, and an app does not have to guess a number; a service that takes one request at a time gives one.

            maximum: 1000 · minimum: 1
          • sleep_after_msinteger | nullrequired

            The most an instance waits without a request before it stops, in milliseconds. Below it, its node picks the wait by the app's own traffic: the wait that would have cost least over the app's last gaps between a request's end and the next one, a second up and idle against a stop and the resume after it in CPU time, so that an app whose requests come closer together than a stop pays for stays up through them, and one whose requests are far apart stops as soon as its last answer has ended (0: stops at once, always). null, which is what it is when nothing is said: the platform's most, three seconds. An app that said 100, the default before 2026-10-28, says null from then on. One that stays stopped is deleted by its node after delete_after_ms, and the next request that needs an instance gets a new one; the instances min_instances keeps awake are neither stopped nor deleted.

            maximum: 600000 · minimum: 0
          • sleep_notice_msinteger | nullrequired

            How long before an instance stops for being idle the programs connected to its guest's /run/sbx/awake are sent the line sleep N, in milliseconds: they flush what they hold, close their connections, or raise a hold, which puts the stop off. The stop comes that long after, whether they answer or not; where nobody is connected, it is not put off. A keyed instance is told as well. 0 or null: not told (the default).

            maximum: 60000 · minimum: 0
          • sleep_hold_max_msintegerrequired

            How long the holds a program in an instance raises keep it up after its time idle has run out, in milliseconds: a program connects to /run/sbx/awake and sends + before it answers a request whose work goes on after the answer, and - (or closes the connection, or dies) when that work is done; the instance stops sleep_after_ms after the last hold is lowered. Past this time it stops whatever holds it, the programs connected told sleep N cap first where sleep_notice_ms asks for a notice. Held, it runs and is metered as running. 0: holds are not heeded. Null: 30000 (the default).

            maximum: 900000 · minimum: 0
          • delete_after_msinteger | nullrequired

            After how long idle an instance is deleted, counted from where its stop is counted from: the next request that needs one gets a new instance, a clone of the version's template, without what the deleted one had gathered in its memory and on its disk (keep what is to last in a volume or a database). A second to an hour, the platform's time, which is the most: more is refused (400 invalid_field), and an instance that is to stay warm is one of min_instances. null, which is what it is when nothing is said: the hour of the nodes (--app-delete-after-seconds). A change holds for the instances the app has as for the ones made after: each node judges every instance by the app's time as it is, once a second at the least. Not for the app's keyed instances, which keys.delete_after_idle_ms is for.

            maximum: 3600000 · minimum: 1000
          • suspend_after_msoptionalinteger | null

            How long a stopped instance stays paused before its memory is saved and it sleeps for real, in milliseconds: paused, the next request is answered in a millisecond or two and its memory is metered; asleep, it wakes in a few more and only what it keeps is. Its node holds it to the longest it holds any instance paused (ten minutes); the host's memory running short, a slot another machine needs and a sleep somebody asks for put it to sleep sooner. Null (the default): its node's longest. A database's suspend_after_seconds is its app's.

            maximum: 86400000 · minimum: 1000
          • wake_daily_atoptionalstring | null

            A time of day in UTC, HH:MM, at which the app's keyed instances and its stateful service's instance that rest are woken every day, for work of their own that waits for a wake (a database's maintenance, say); the instances the nodes make and delete by the app's requests are woken by requests alone. The wake is no activity: what runs then is, and once that is done the instance rests again. A stateful service's instance is woken besides for the next run of its PostgreSQL server's pg_cron jobs, whatever this says. Null (the default): none. A database's wake_daily_at is its app's.

            pattern: ^([01][0-9]|2[0-3]):[0-5][0-9]$
        • keysrequired

          Una de estas opciones:

          • object
          • null
        • egressobject | nullrequired

          Egress rules as a sandbox has them (default, allow, deny, http, inject), null when there are none; kept as the calls gave them (a PATCH merges them key by key, as a node does). They hold at once and make no version: the nodes that serve the app put them on every instance of its group, the ones there are (one that runs for the connections it has too) and the ones made after; a keyed instance does not have them. The control plane checks only that they are an object of those keys of at most 256 KiB: what the rules say is a node's to check, when it applies them, and an instance whose rules its node cannot read is refused everything (the node's log says why).

          • defaultoptionalallowodenyonull
          • allowoptionalarray | null
          • denyoptionalarray | null
          • httpoptionalarray | null
          • injectoptionalarray | null
          • deny_portsoptionalarray | null

            Ports no connection reaches, TCP or UDP, wherever it goes and whatever else the rules say. A Browsers session's always hold 25, 465, 587 and the ports mining pools listen on (3333, 4444, 5555, 7777, 14444, 45700): the control plane adds them to whatever a change of its rules gives.

        • metadatamapa destringrequired
        • logsobjectrequired

          What the app's log has, each part as it was given or at its default (AppLogSettings).

          • outputbooleanrequired
          • keepbooleanrequired
          • requestsobjectrequired
            • sample_ratenumberrequired
              maximum: 1 · minimum: 0
            • keep_errorsbooleanrequired
        • publicbooleanrequired

          Whether it answers at its default domain (url).

        • depends_onarray deobjectrequired
          • servicestringrequired
          • conditionstringrequired
        • networkobjectrequired
          • aliasesarray destringrequired
          • endpoint_modestringrequired
        • drivesoptionalarray deobject

          The Drives its instance mounts (AppCreate.drives): none, [], where it names none.

          • drivestringrequired
          • pathstringrequired
          • modestringrequired
        • key_drivesoptionalarray deobject

          The Drives each key's instance has of its own (AppCreate.key_drives): none, [], where it gives none.

          • namestringrequired
          • pathstringrequired
          • size_bytesinteger | nullrequired
            format: int64
          • nocopyoptionalunknown

            There only where it was given as true.

        • runtimeoptional

          Its runtime settings (AppCreate.runtime). null: none.

          Una de estas opciones:

          • Runtime
            • ulimitsmapa deRuntimeLimitGivenrequired
            • sysctlsmapa destringrequired
            • tmpfsarray deobjectrequired
              • pathstringrequired
              • size_bytesinteger | nullrequired
              • modeintegerrequired
            • dnsobjectrequired
              • serversarray destringrequired
              • searcharray destringrequired
              • optionsarray destringrequired
            • shm_size_bytesinteger | nullrequired
          • null
        • tracesoptionalobject

          What the app's requests' traces have, each part as it was given or at its default (AppTraceSettings): no span where sample_rate is 0.

          • sample_ratenumberrequired
            maximum: 1 · minimum: 0
          • trust_incomingbooleanrequired
    • volumesarray deobjectrequired

      The volumes of the project the services mount.

      • modelstring | nullrequired

        Its name in the model. Null: made for a path a stateful service mounts with no volume named.

      • namestringrequired

        Its name in the project.

      • idstring | nullrequired

        Null: not made yet (to_make).

        pattern: ^vol-
      • sharedbooleanrequired

        Several services mount it, or one with several instances: every instance of each mounts it at once.

      • statusto_makeomadeothereoexternalrequired

        to_make: the import makes it (a dry run says so of what it would make); made: this import made it; there: an earlier import made it, or a volume of its name was there; external: the model names it external, and the project has it.

      • used_byarray destringrequired
    • secretsarray deobjectrequired

      The secrets of the model that services are granted.

      • modelstringrequired

        Its name in the model.

      • namestringrequired

        The project's secret that holds it: the model's name, with each character a secret's name does not take written as an underscore.

      • therebooleanrequired

        Whether the project has it. An import (not a dry run) with one that is not there is refused 404 secret_not_found.

      • used_byarray destringrequired
    • left_outarray deobjectrequired

      The services of profiles that are not turned on, which the import does not touch.

      • namestringrequired
      • profilesarray destringrequired
    • absentarray destringrequired

      The project's services that the model does not have: kept as they are.

    • deploystring | nullrequired

      The deploy the import made (GET /v1/deploys/{id}). Null: a dry run, or nothing would get a version.

      pattern: ^dpl-
  • accountsoptionalarray deobject

    With sole_owner: the accounts the user is the only owner of.

    • idstringrequired
      pattern: ^acct-[0-9a-hjkmnp-tv-z]{16}$
    • namestringrequired

409

name_taken: another app has the name, or has it as an alias, or has an alias as a name or an alias (param is name or network.aliases[<n>]); volumes_apart; app_changed: other calls changed the app while this one was worked out, again and again (retryable). drive_single_writer: drives of an app of many instances, or of keys: a Drive has one writer (param names what makes it many; an app's version mounts no snapshot of a Drive read-only yet); too_many_drives: more than 8 Drives and volumes of its own an instance would mount; key_drive: drives names a key's Drive. managed_by_database: the app is a database's (managed_by), changed through /v1/databases alone. managed_by_sandboxes: the app is the Sandboxes product's (managed_by), whose instances are the project's sandboxes, made, changed and deleted through /v1/sandboxes.
JSON
  • errorobjectrequired
    • typestringrequired
    • codestringrequired
    • messagestringrequired
    • paramstring | nullrequired
    • retryablebooleanrequired
    • request_idstringrequired
    • docs_urlstring | nullrequired
  • move_idoptionalstring

    With already_moving: the move of the sandbox that is under way.

  • upgrade_idoptionalstring

    With upgrade_in_progress (the key's upgrade that is not over), upgrade_started (the upgrade that has begun) and rollback_gone (the upgrade whose revert is refused): that upgrade.

    pattern: ^upg-[0-9a-hjkmnp-tv-z]{16}$
  • moveoptionalobject

    With move_over, move_going_back and move_taken_over (POST /internal/moves/{id}/cancel): the move, as it stands.

    • objectstringrequired
    • idstringrequired
      pattern: ^mov-[0-9a-hjkmnp-tv-z]{16}$
    • sandbox_idstringrequired
    • project_idstringrequired
    • fromobjectrequired
      • idstringrequired
      • namestring | nullrequired

        Its name; null for a node that is off the books since.

    • toobjectrequired
      • idstringrequired
      • namestring | nullrequired

        Its name; null for a node that is off the books since.

    • epochintegerrequired

      The epoch the table held the sandbox at when the move began: the old node releases it at it, the new one takes it over one on, the old one takes it back two on, and from the object store three on.

      minimum: 1
    • stepstringrequired

      Where it stands: releasing (the old node is asked to release it; the table has it there), taking_over (the new node is asked to take it over; the table has it there), reclaiming (the new node refused it, and the old one is asked to take it back), restoring (the old one kept nothing of it, and takes it over from the object store), done (on the new node), failed (where it was, or back there, or where a failure no time clears or an operator's cancel found it: error says why).

    • whystringrequired

      What made it: the operator's move, an evacuation, a recovery.

    • wakebooleanrequired

      Whether the new node wakes it: as the call said, or as it was (it ran) when the move began.

    • from_storebooleanrequired

      The new node takes it from the object store alone.

    • releaseobject | nullrequired

      What the old node's release answered (sandboxd/openapi/openapi.json, releaseSandbox), without the release's key; null before it answered.

    • takeoverobject | nullrequired

      What the takeover that ended it answered (takeoverSandbox of the node's document): where its disk and memory came from (source), how long each took (ms), memory; null before.

    • errorrequired

      Why it failed, of one that failed (the new node's refusal, where it was taken back to the old node); null, of one that is done; of one under way, what the step it is at was last told where it is asked again.

      Una de estas opciones:

      • object
      • null
    • attemptsintegerrequired

      How many times the step it is at was asked without an answer that settles it.

      minimum: 0
    • next_attempt_atstring | nullrequired

      When that step is asked again; null where it is not waiting.

      format: date-time
    • batchstring | nullrequired

      The evacuation (evc-…) or recovery (rcv-…) it is of.

    • created_atstringrequired
      format: date-time
    • updated_atstringrequired
      format: date-time
    • ended_atstring | nullrequired
      format: date-time
  • lease_untiloptionalstring

    With lease_held: until when the node may still hold the lease it was granted last, and the time after it for the node to have seen it run out (recoverNode is refused until then).

    format: date-time
  • activeoptionalobject | null

    With standby: the active copy, as the standby last read the lease; null while no copy holds it.

    • namestringrequired
    • urlstringrequired
    • termintegerrequired
      minimum: 1
  • allowed_hostsoptionalarray | null

    With secret_host_not_allowed: the hosts the secret may go to, as the secret has them; null where it was given none (the node's SecretHosts).

  • problemsoptionalarray deobject

    With invalid_definition: every problem of the definition, each where it is and what it is.

    • atstringrequired

      Where in the definition, a dotted path (services.web.port); empty for the whole of it.

    • messagestringrequired
  • reportoptionalobject

    With compose_refused: the import's report, which says what is refused, why, and what to change.

    • objectcompose_importrequired
    • namestring | nullrequired

      The model's name: each service's label compose.project. Null: it has none.

    • dry_runbooleanrequired
    • strictbooleanrequired
    • profilesarray destringrequired
    • treatmentsobjectrequired

      How many items of the report have each treatment.

      • doneintegerrequired
        minimum: 0
      • adaptedintegerrequired
        minimum: 0
      • ignoredintegerrequired
        minimum: 0
      • refusedintegerrequired
        minimum: 0
    • itemsarray deobjectrequired

      The model's top-level elements: name, each volume, network, config and secret, extensions.

      • atstringrequired

        Where in the model: of a service, its attribute (ports[0], depends_on.db, environment.KEY), empty for the service as a whole; of the top level, the element (volumes.db-data).

      • treatmentdoneoadaptedoignoredorefusedrequired

        done: it means the same on the platform; adapted: the platform does the same another way; ignored: it matters to a local run alone (refused with strict); refused: the platform cannot do it, and doing without it would let the program go wrong quietly.

      • whystringrequired
      • changestring | nullrequired

        Of one that is refused: what to change. Null otherwise.

    • servicesarray deobjectrequired

      The services the import takes, in the model's order.

      • namestringrequired

        The service's name, which is its app's.

      • app_idstring | nullrequired

        Its app. Null: a dry run of a service the project does not have.

        pattern: ^app-
      • statuscreatedochangedounchangedrequired

        What the import does (or, in a dry run, would do) to its app: makes it, writes what changed, or nothing.

      • kindserviceojobrequired

        job: a service of the model waits for it to complete, and it runs to its end in each deploy that needs it.

      • statefulbooleanrequired

        It mounts a volume of its own: one instance, which has the volume.

      • versionbooleanrequired

        Whether the import's deploy makes a version of it: what its version would be made of, its references resolved, is not what its current version is.

      • itemsarray deobjectrequired
        • atstringrequired

          Where in the model: of a service, its attribute (ports[0], depends_on.db, environment.KEY), empty for the service as a whole; of the top level, the element (volumes.db-data).

        • treatmentdoneoadaptedoignoredorefusedrequired

          done: it means the same on the platform; adapted: the platform does the same another way; ignored: it matters to a local run alone (refused with strict); refused: the platform cannot do it, and doing without it would let the program go wrong quietly.

        • whystringrequired
        • changestring | nullrequired

          Of one that is refused: what to change. Null otherwise.

      • appobjectrequired

        The fields of the service's app as the import writes them: what its versions are made of, and what holds at once.

        • imagestring | nullrequired

          What the app's next version is made of, with the seven fields after it: its newest version's, with what was changed since. null: the app has no image, and no version.

        • image_pull_secretstring | nullrequired

          The name of the secret that holds the credentials of the image's registry. No answer has its value. null: none.

        • start_commandrequired

          Una de estas opciones:

          • StartCommand
          • null
        • start_argsrequired

          What follows the image's ENTRYPOINT in the place of its CMD. null: the image's CMD, or the whole of start_command.

          Una de estas opciones:

          • array<string>
          • null
        • portinteger | nullrequired
        • readyobjectrequired

          How to tell that an instance is up, which a version's template is saved after: exactly one of the three. The first two are asked from outside the instance, so the service has to listen on every address (0.0.0.0).

          • httpoptionalstring

            A path, beginning with /: a GET of it on the app's port is answered with a status from 200 to 399.

          • tcpoptionaltrue

            The app's port takes a connection.

          • commandoptional

            A command that succeeds in the instance. It is run by the instance's shell: a list of arguments is written as one line for it.

            Una de estas opciones:

            • string
            • array<string>
        • envmapa destringrequired
        • secretsmapa destringrequired

          Names of secrets. No answer has a value.

        • sizeobjectrequired

          What one instance may use at most. It is billed by what it uses.

          • vcpusintegerrequired
            minimum: 1
          • memory_bytesintegerrequired

            An even number of MiB (a whole number of the 2 MiB pages a node loads a machine's memory in), in bytes. Any other size is 400 invalid_field (param size.memory_bytes): no node would make an instance of it, as none makes a sandbox of it (the node's SandboxLimitsCreate.memory_bytes).

            minimum: 134217728
        • volumesarray deobjectrequired
          • volumestringrequired
          • pathstringrequired
          • read_onlybooleanrequired
          • nocopyoptionalunknown

            There only where it was given as true.

        • kindstringrequired
        • filesarray deobjectrequired
          • pathstringrequired
          • modeintegerrequired
          • uidintegerrequired
          • gidintegerrequired
          • contentoptionalstring
          • secretoptionalstring
        • userstring | nullrequired
        • working_dirstring | nullrequired
        • stoprequired

          Una de estas opciones:

          • ServiceStop
            • signalstring | nullrequired
            • grace_msinteger | nullrequired
          • null
        • hostnamestring | nullrequired
        • regionsarray destringrequired

          In the order na, eu, ap.

        • scaleobjectrequired
          • min_instancesintegerrequired

            How many instances of the app are kept awake in a region, whether a request comes or not: the floor. The nodes of the region that serve the app share it as they share max_instances (two nodes and a floor of 2: one on each; a floor of 1: one, on the node that is tried first; a floor of 5: three and two), and a node that is away is made up for by the others. A node makes its part when it is told, not at the first request, and neither stops nor deletes those instances for being idle; they are metered as instances that are awake. 0, which is what it is when nothing is said: none. Lowered, the instances beyond it are idle from that moment, and are stopped and deleted as idle instances are. At most max_instances, where that is given, and at most what the account's plan lets an app ask for in a region (caps.app_instances.limit: 5, 50 and 500): more is refused, 403 plan_cap_reached.

            minimum: 0
          • max_instancesinteger | nullrequired

            How many instances the app has at the most in a region: the nodes of the region that serve the app share it. At most what the account's plan lets an app ask for (caps.app_instances.limit: 5 on Free, 500 on Paid): more is refused, 403 plan_cap_reached, and what an app asked for on a plan that let it is kept, and held to the plan's number. null: the plan's number for an app that names none (caps.app_instances.default: 2 on Free, 50 on Paid), or min_instances where that is more. What the nodes are told is no more than the account has room for awake beside everything else of it that is awake, nor less than the app has awake.

            minimum: 1
          • concurrencyinteger | nullrequired

            A cap on what one instance has in hand at a time: the requests the edges sent it, and the connections it has open on its project's network (through its service's address, which the node shares out among the instances connection by connection, or to its own). An instance with that many is full, and another is made for more. null, which is what it is when nothing is said: no cap. A node then adds an instance when the ones it has are near their size in CPU or memory, and an app does not have to guess a number; a service that takes one request at a time gives one.

            maximum: 1000 · minimum: 1
          • sleep_after_msinteger | nullrequired

            The most an instance waits without a request before it stops, in milliseconds. Below it, its node picks the wait by the app's own traffic: the wait that would have cost least over the app's last gaps between a request's end and the next one, a second up and idle against a stop and the resume after it in CPU time, so that an app whose requests come closer together than a stop pays for stays up through them, and one whose requests are far apart stops as soon as its last answer has ended (0: stops at once, always). null, which is what it is when nothing is said: the platform's most, three seconds. An app that said 100, the default before 2026-10-28, says null from then on. One that stays stopped is deleted by its node after delete_after_ms, and the next request that needs an instance gets a new one; the instances min_instances keeps awake are neither stopped nor deleted.

            maximum: 600000 · minimum: 0
          • sleep_notice_msinteger | nullrequired

            How long before an instance stops for being idle the programs connected to its guest's /run/sbx/awake are sent the line sleep N, in milliseconds: they flush what they hold, close their connections, or raise a hold, which puts the stop off. The stop comes that long after, whether they answer or not; where nobody is connected, it is not put off. A keyed instance is told as well. 0 or null: not told (the default).

            maximum: 60000 · minimum: 0
          • sleep_hold_max_msintegerrequired

            How long the holds a program in an instance raises keep it up after its time idle has run out, in milliseconds: a program connects to /run/sbx/awake and sends + before it answers a request whose work goes on after the answer, and - (or closes the connection, or dies) when that work is done; the instance stops sleep_after_ms after the last hold is lowered. Past this time it stops whatever holds it, the programs connected told sleep N cap first where sleep_notice_ms asks for a notice. Held, it runs and is metered as running. 0: holds are not heeded. Null: 30000 (the default).

            maximum: 900000 · minimum: 0
          • delete_after_msinteger | nullrequired

            After how long idle an instance is deleted, counted from where its stop is counted from: the next request that needs one gets a new instance, a clone of the version's template, without what the deleted one had gathered in its memory and on its disk (keep what is to last in a volume or a database). A second to an hour, the platform's time, which is the most: more is refused (400 invalid_field), and an instance that is to stay warm is one of min_instances. null, which is what it is when nothing is said: the hour of the nodes (--app-delete-after-seconds). A change holds for the instances the app has as for the ones made after: each node judges every instance by the app's time as it is, once a second at the least. Not for the app's keyed instances, which keys.delete_after_idle_ms is for.

            maximum: 3600000 · minimum: 1000
          • suspend_after_msoptionalinteger | null

            How long a stopped instance stays paused before its memory is saved and it sleeps for real, in milliseconds: paused, the next request is answered in a millisecond or two and its memory is metered; asleep, it wakes in a few more and only what it keeps is. Its node holds it to the longest it holds any instance paused (ten minutes); the host's memory running short, a slot another machine needs and a sleep somebody asks for put it to sleep sooner. Null (the default): its node's longest. A database's suspend_after_seconds is its app's.

            maximum: 86400000 · minimum: 1000
          • wake_daily_atoptionalstring | null

            A time of day in UTC, HH:MM, at which the app's keyed instances and its stateful service's instance that rest are woken every day, for work of their own that waits for a wake (a database's maintenance, say); the instances the nodes make and delete by the app's requests are woken by requests alone. The wake is no activity: what runs then is, and once that is done the instance rests again. A stateful service's instance is woken besides for the next run of its PostgreSQL server's pg_cron jobs, whatever this says. Null (the default): none. A database's wake_daily_at is its app's.

            pattern: ^([01][0-9]|2[0-3]):[0-5][0-9]$
        • keysrequired

          Una de estas opciones:

          • object
          • null
        • egressobject | nullrequired

          Egress rules as a sandbox has them (default, allow, deny, http, inject), null when there are none; kept as the calls gave them (a PATCH merges them key by key, as a node does). They hold at once and make no version: the nodes that serve the app put them on every instance of its group, the ones there are (one that runs for the connections it has too) and the ones made after; a keyed instance does not have them. The control plane checks only that they are an object of those keys of at most 256 KiB: what the rules say is a node's to check, when it applies them, and an instance whose rules its node cannot read is refused everything (the node's log says why).

          • defaultoptionalallowodenyonull
          • allowoptionalarray | null
          • denyoptionalarray | null
          • httpoptionalarray | null
          • injectoptionalarray | null
          • deny_portsoptionalarray | null

            Ports no connection reaches, TCP or UDP, wherever it goes and whatever else the rules say. A Browsers session's always hold 25, 465, 587 and the ports mining pools listen on (3333, 4444, 5555, 7777, 14444, 45700): the control plane adds them to whatever a change of its rules gives.

        • metadatamapa destringrequired
        • logsobjectrequired

          What the app's log has, each part as it was given or at its default (AppLogSettings).

          • outputbooleanrequired
          • keepbooleanrequired
          • requestsobjectrequired
            • sample_ratenumberrequired
              maximum: 1 · minimum: 0
            • keep_errorsbooleanrequired
        • publicbooleanrequired

          Whether it answers at its default domain (url).

        • depends_onarray deobjectrequired
          • servicestringrequired
          • conditionstringrequired
        • networkobjectrequired
          • aliasesarray destringrequired
          • endpoint_modestringrequired
        • drivesoptionalarray deobject

          The Drives its instance mounts (AppCreate.drives): none, [], where it names none.

          • drivestringrequired
          • pathstringrequired
          • modestringrequired
        • key_drivesoptionalarray deobject

          The Drives each key's instance has of its own (AppCreate.key_drives): none, [], where it gives none.

          • namestringrequired
          • pathstringrequired
          • size_bytesinteger | nullrequired
            format: int64
          • nocopyoptionalunknown

            There only where it was given as true.

        • runtimeoptional

          Its runtime settings (AppCreate.runtime). null: none.

          Una de estas opciones:

          • Runtime
            • ulimitsmapa deRuntimeLimitGivenrequired
            • sysctlsmapa destringrequired
            • tmpfsarray deobjectrequired
              • pathstringrequired
              • size_bytesinteger | nullrequired
              • modeintegerrequired
            • dnsobjectrequired
              • serversarray destringrequired
              • searcharray destringrequired
              • optionsarray destringrequired
            • shm_size_bytesinteger | nullrequired
          • null
        • tracesoptionalobject

          What the app's requests' traces have, each part as it was given or at its default (AppTraceSettings): no span where sample_rate is 0.

          • sample_ratenumberrequired
            maximum: 1 · minimum: 0
          • trust_incomingbooleanrequired
    • volumesarray deobjectrequired

      The volumes of the project the services mount.

      • modelstring | nullrequired

        Its name in the model. Null: made for a path a stateful service mounts with no volume named.

      • namestringrequired

        Its name in the project.

      • idstring | nullrequired

        Null: not made yet (to_make).

        pattern: ^vol-
      • sharedbooleanrequired

        Several services mount it, or one with several instances: every instance of each mounts it at once.

      • statusto_makeomadeothereoexternalrequired

        to_make: the import makes it (a dry run says so of what it would make); made: this import made it; there: an earlier import made it, or a volume of its name was there; external: the model names it external, and the project has it.

      • used_byarray destringrequired
    • secretsarray deobjectrequired

      The secrets of the model that services are granted.

      • modelstringrequired

        Its name in the model.

      • namestringrequired

        The project's secret that holds it: the model's name, with each character a secret's name does not take written as an underscore.

      • therebooleanrequired

        Whether the project has it. An import (not a dry run) with one that is not there is refused 404 secret_not_found.

      • used_byarray destringrequired
    • left_outarray deobjectrequired

      The services of profiles that are not turned on, which the import does not touch.

      • namestringrequired
      • profilesarray destringrequired
    • absentarray destringrequired

      The project's services that the model does not have: kept as they are.

    • deploystring | nullrequired

      The deploy the import made (GET /v1/deploys/{id}). Null: a dry run, or nothing would get a version.

      pattern: ^dpl-
  • accountsoptionalarray deobject

    With sole_owner: the accounts the user is the only owner of.

    • idstringrequired
      pattern: ^acct-[0-9a-hjkmnp-tv-z]{16}$
    • namestringrequired

Escribe para buscar…

↑↓ navegar↵ seleccionarEsc cerrar