Record why a deployment has no pack service yet
const url = 'https://example.com/v1/system/tenants/2489E9AD-2EE2-8E00-8EC9-32D5F69181C0/deployments/example/endpoint/problem';const options = { method: 'PUT', headers: {Authorization: 'Bearer <token>', 'Content-Type': 'application/json'}, body: '{"code":"endpoint-url-invalid"}'};
try { const response = await fetch(url, options); const data = await response.json(); console.log(data);} catch (error) { console.error(error);}curl --request PUT \ --url https://example.com/v1/system/tenants/2489E9AD-2EE2-8E00-8EC9-32D5F69181C0/deployments/example/endpoint/problem \ --header 'Authorization: Bearer <token>' \ --header 'Content-Type: application/json' \ --data '{ "code": "endpoint-url-invalid" }'For whatever creates the pack services: records why a deployment waiting for one has none yet (pending, or cleared and waiting, as after a suspension is lifted), as one of the fixed codes (tenant-not-provisioned, capacity-unavailable, service-start-failed, image-not-allowed, and the endpoint checks’). The tenant sees the code and when, so a deploy waiting for ready can say why at once. Takes system:deployments.
Authorizations
Section titled “Authorizations”Parameters
Section titled “Parameters”Path Parameters
Section titled “Path Parameters”The tenant’s id (a UUID).
The deployment’s id, as GET /v1/system/deployments lists it.
Header Parameters
Section titled “Header Parameters”Caller-supplied idempotency key, scoped to the caller: the same key from someone else in the workspace is their own request (from 0.1.5). Retries with the same key return the original response byte-identical (per docs/API-ROUTE-CONVENTIONS.md §3.1). A retry sent while the first request still runs is answered 409 idempotency-key-in-flight with Retry-After, on a runtime whose store holds keys (from 0.1.5); retry after it, and you get the first request’s answer. An answer that carries a secret (a new API key or public run token, a session token, a generated signing secret) isn’t kept: a retry gets 409 idempotency-key-replay-withheld, with the first request’s status and when it succeeded (from 0.1.5). A request whose key store can’t be read before it runs is answered 503 idempotency-store-unavailable with Retry-After: nothing ran, so retry with the same key (from 0.3.0).
Request Bodyrequired
Section titled “Request Bodyrequired”object
Why the pack service couldn’t be created, or was refused: one of DEPLOYMENT_ENDPOINT_PROBLEM_CODES. The tenant sees the code, never more.
Responses
Section titled “Responses”Recorded: the deployment’s endpoint as its tenant sees it.
Its pack service’s status, when the server routes each pack call to the deployment that shipped the tool (KINDGI_PACK_ROUTING=per-deployment). Absent otherwise, and for a deployment that ships nothing a pack service serves (no tool, and no guardrail check but built-in ones: a pack of agents and flows only), which needs none. pending: no pack service yet, and calls to its tools fail (pack-endpoint-pending) rather than go to another deployment. ready: calls go to its own pack service. cleared: its pack service was taken down, and calls fail (pack-endpoint-cleared); it carries waiting or retired to say whether it comes back. Carries no URL and no token.
object
When the status last changed.
While the deployment waits for a pack service (pending, or cleared and waiting): the last reason recorded while it waited. A ready endpoint has none, nor a retired one. A fixed code, never free text.
object
One of a fixed list; a later runtime may add codes, so treat an unknown one as a reason not yet named. The server refused the endpoint it was given: endpoint-url-invalid (not https, or with user info or a path), endpoint-host-not-allowed (the host isn’t on the operator’s allowlist), endpoint-name-mismatch (the host doesn’t name this deployment’s service), endpoint-unreachable (the pack service didn’t answer), endpoint-info-mismatch (it answered as another pack or version), image-not-allowed (the image is outside the workspace’s repository). The pack service couldn’t be created: tenant-not-provisioned (the workspace has nowhere to run pack services yet), capacity-unavailable (no room for another pack service just now), service-start-failed (the image was deployed but didn’t start). It can’t be given its declared environment (GET /v1/deployments/{deploymentId}/pack-env names the names): pack-env-ambiguous, pack-env-secret-not-yet, pack-env-missing, pack-env-secret-unreadable (a name’s secret can’t be read for a reason the platform fixes, not the owner; it’s tried again). It can’t be given its secrets: pack-env-secrets-unaudited (the runtime has nowhere to audit a secret it gives out; the platform fixes it, and it’s tried again), pack-env-secret-too-large (a secret’s value is over 32768 bytes, the most a pack service can be given), pack-env-name-too-long (a name set as a secret is too long for the secret it’s kept in, which is named after the pack service; shorten it in the pack), pack-env-secret-capacity (a name is set only as a secret, and there’s no room just now for a pack service that can be given one, apart from other projects’; it’s tried again).
When it was recorded.
cleared only: a pack service is wanted for it again, as once a suspension that took its service down is lifted, and it lists as needing one; problem says why it isn’t back yet. A cleared endpoint carries this or retired, never both. Absent from older servers, which send neither.
cleared only: why no pack service is wanted for it now. superseded: nothing a call can still ask for needs it: for each tool version, pack check, agent and flow version it serves, a newer deployment into its project serves it too, or it’s unregistered and not held (a check: no registered guardrail runs it in its project); and each tool it deployed without a version, a newer deployment into its project deploys without one too (from Kindgi 0.4.0; before, it meant newer deployments into its project shipped every tool version and check it does and deployed each tool it deployed without a version the same way, and was for good). One whose endpoint an earlier runtime made or last set is superseded as well, for good, once newer deployments into its project ship every tool version and check it does and deploy each tool it deployed without a version the same way. Otherwise it’s needed again once something it serves that no newer deployment into its project serves is wanted again: a tool, agent or flow version registered or held again, or a check a newly registered guardrail runs in its project. A tool call a step made on it still goes back to it when the step runs again, and fails closed (pack-endpoint-cleared). tenant-closing, tenant-suspended, project-suspended and pack-suspended last until the workspace, project or pack is restored. A later runtime may add reasons, so treat an unknown one as a reason not yet named. A cleared endpoint carries this or waiting, never both. Absent from older servers, which send neither.
ready only: the pack service runs on the values it was last given, because the newest can’t be given to it (a required one was removed, say, or a secret can’t be given), or didn’t start (pack-env-update-failed). Gone once they can, or do. For a name that can’t be given, GET /v1/deployments/{deploymentId}/pack-env names it. Absent from older servers.
object
Why a ready pack service runs on the values it was last given: a PackEnvProblemCode, so the newest can’t be given to it; pack-env-update-failed: they were given, and its new revision with them didn’t start, so the last ready one serves. That revision is tried again for an hour after it’s first recorded; setting the values again is a new revision, tried again too. The orchestrator couldn’t give it its secrets: pack-env-secrets-unaudited (the runtime has nowhere to audit a secret it gives out, so it gives none; the platform fixes it), pack-env-secret-too-large (a secret’s value is over 32768 bytes, the most a pack service can be given), pack-env-name-too-long (a name set as a secret is too long for the secret it’s kept in, which is named after the pack service; shorten it in the pack), pack-env-secret-redeploy (a name is now set only as a secret, and this pack service was made where no secret can be given to it; a new deployment’s can be, from a new build of the pack), pack-env-cloud-secret-unsupported (a name is set to a cloud secret, and this pack service’s host mounts none; removing it, or setting a value instead, moves the revision). A service held as pack-env-secret-unreadable or pack-env-secrets-unaudited is listed by GET /v1/system/pack-env every time, with its held, as the revision doesn’t move when the platform fixes it. A later runtime may add codes, so treat an unknown one as a reason not yet named.
When it was recorded.
What keeps this build’s pack service, whatever its status, in kind, id, then version order. [] when nothing does: it’s retired (superseded), or retired next. A suspended build has retired and still lists what keeps it, since its service comes back when the suspension lifts. On a list page, the first 20 entries, with keptForTotal; a get of one deployment gives them all. Absent when builds aren’t counted (one service serves every build), and from a runtime before Kindgi 0.4.0.
One thing a build’s pack service is kept for. To let the build go, unregister it: kindgi tools|agents|flows unregister <id> <version>, kindgi guardrails unregister <id>.
object
tool: a tool version this build is the newest into its project to route, still registered or held; a tool deployed without a version has no version and neither flag, and is kept until a newer deployment into the project ships the tool. agent, flow: a version this build is the newest into its project to ship or carry, still registered or held. guardrail: a registered guardrail that runs a pack check this build is the newest into its project to route; it has no version. A later runtime may add kinds, so treat an unknown one as a kind not yet named.
The tool, agent, flow or guardrail id.
The version kept for. Absent for a guardrail, or a tool deployed without one.
The version (or guardrail) is registered.
A pin or a run that hasn’t finished uses the version. Unregistering it then lets the build go only once the pin’s own version is unregistered or the run finishes. Sent beside registered when both are true.
On a list page, with keptFor: how many entries there are, of which keptFor holds the first 20. A get of one deployment doesn’t send it.
Example
{ "status": "pending", "waiting": true, "packEnvProblem": { "code": "pack-env-ambiguous" }, "keptFor": [ { "registered": true, "held": true } ]}A malformed body, or a code not on the list (bad-input).
object
object
Stable machine-readable discriminant. Values match domain error codes (see docs/API-ROUTE-CONVENTIONS.md §4.3).
Optional, kind-specific.
object
Server-assigned request id; also echoed via X-Request-Id header.
Examplegenerated
{ "error": { "code": "example", "message": "example", "details": {}, "requestId": "example" }}Missing / malformed / expired / revoked bearer token.
object
object
Stable machine-readable discriminant. Values match domain error codes (see docs/API-ROUTE-CONVENTIONS.md §4.3).
Optional, kind-specific.
object
Server-assigned request id; also echoed via X-Request-Id header.
Examplegenerated
{ "error": { "code": "example", "message": "example", "details": {}, "requestId": "example" }}A tenant’s key or session (system-principal-required), or a system key without system:deployments (system-scope-missing).
object
object
Stable machine-readable discriminant. Values match domain error codes (see docs/API-ROUTE-CONVENTIONS.md §4.3).
Optional, kind-specific.
object
Server-assigned request id; also echoed via X-Request-Id header.
Examplegenerated
{ "error": { "code": "example", "message": "example", "details": {}, "requestId": "example" }}No such tenant (tenant-not-found), or no such deployment in it (deployment-not-found).
object
object
Stable machine-readable discriminant. Values match domain error codes (see docs/API-ROUTE-CONVENTIONS.md §4.3).
Optional, kind-specific.
object
Server-assigned request id; also echoed via X-Request-Id header.
Examplegenerated
{ "error": { "code": "example", "message": "example", "details": {}, "requestId": "example" }}The deployment’s endpoint isn’t waiting for a pack service: ready, or cleared and not waiting (endpoint-not-pending, with details.status).
object
object
Stable machine-readable discriminant. Values match domain error codes (see docs/API-ROUTE-CONVENTIONS.md §4.3).
Optional, kind-specific.
object
Server-assigned request id; also echoed via X-Request-Id header.
Examplegenerated
{ "error": { "code": "example", "message": "example", "details": {}, "requestId": "example" }}The request body is over the API’s limit (request-too-large; details.maxBytes says how much).
object
object
Stable machine-readable discriminant. Values match domain error codes (see docs/API-ROUTE-CONVENTIONS.md §4.3).
Optional, kind-specific.
object
Server-assigned request id; also echoed via X-Request-Id header.
Examplegenerated
{ "error": { "code": "example", "message": "example", "details": {}, "requestId": "example" }}Server error (unmapped domain code or framework crash).
object
object
Stable machine-readable discriminant. Values match domain error codes (see docs/API-ROUTE-CONVENTIONS.md §4.3).
Optional, kind-specific.
object
Server-assigned request id; also echoed via X-Request-Id header.
Examplegenerated
{ "error": { "code": "example", "message": "example", "details": {}, "requestId": "example" }}The operator’s access record can’t be written, so nothing was done (access-record-unavailable), or the Idempotency-Key store couldn’t be read before the request ran (idempotency-store-unavailable, with Retry-After).
object
object
Stable machine-readable discriminant. Values match domain error codes (see docs/API-ROUTE-CONVENTIONS.md §4.3).
Optional, kind-specific.
object
Server-assigned request id; also echoed via X-Request-Id header.
Examplegenerated
{ "error": { "code": "example", "message": "example", "details": {}, "requestId": "example" }}See also
GET /v1/system/tenants: List tenantsPOST /v1/system/tenants: Create a tenantGET /v1/system/tenants/{tenantId}: Read a tenantPATCH /v1/system/tenants/{tenantId}: Rename a tenant- Error code
access-record-unavailable - Error code
bad-input - Error code
deployment-not-found - Error code
endpoint-not-pending - Error code
idempotency-store-unavailable - Error code
request-too-large - Error code
system-principal-required - Error code
system-scope-missing - Error code
tenant-not-found - Every error code