enumeray connector grant update

Change a compute grant’s limits while retaining its identity and publications.

Use when: Routine capacity changes should apply without revocation or republishing.

Role and prerequisites: Connector owner or current co-admin. Requires connectors:manage. Install enumeray, configure the API and log in. Use an account or delegated token with the required permissions. Example IDs must be replaced with IDs from your service.

Example

enumeray connector grant get lab-connector 33333333-3333-4333-8333-333333333333
enumeray connector grant update lab-connector 33333333-3333-4333-8333-333333333333 --max-queued-runs 50 --max-outstanding-assignments 2 --generation 1

Resource names and UUIDs below are examples. Replace them with IDs returned by your service; --generation 1 is valid only if the latest response reports generation 1. Both limits must be supplied. A stale generation returns 409, including an identical retry after a committed update whose reply was lost. Inspect the grant and audit before deliberately applying another update. The CLI never refreshes the generation or retries automatically. Revoked grants and archived connector/contexts cannot be updated; a draining context stays draining. An unchanged request at the current generation does not increment or audit again.

Inputs and options

Input	Meaning and use
<CONNECTOR>	Owning connector ID.
<GRANT_ID>	Compute-grant ID returned by connector grant create or listed by connector grant list; this is not a context ID.
--max-queued-runs <MAX_QUEUED_RUNS>	Required new grant-wide queued-run limit, 1–10000. Includes standard and candidate trial runs across publications and models; accepted queued runs remain after a decrease.
--max-outstanding-assignments <MAX_OUTSTANDING_ASSIGNMENTS>	Required new grant-wide assignment limit, 1–1024. Preparing, scheduler-waiting, executing and collecting/uploading assignments occupy capacity until terminal reporting; uncertain work retains capacity.
--generation <GENERATION>	Observed grant generation from connector grant get or list; distinct from context, connector configuration and ownership generations.

Every command also accepts the profile, API, JSON and timeout options.

Result and next step

Changes both limits atomically and returns the current grant record and generation. Existing queued and assigned work remains. At the new limit, fresh submissions return HTTP 429 and fresh claims return HTTP 204; capacity becomes available as usage drains below the limit.

Inspect the resulting limits and existing connector diagnostics.

Exact syntax, defaults and accepted values

Replace both capacity limits at the observed grant generation

Usage: enumeray connector grant update [OPTIONS] --max-queued-runs <MAX_QUEUED_RUNS> --max-outstanding-assignments <MAX_OUTSTANDING_ASSIGNMENTS> --generation <GENERATION> <CONNECTOR> <GRANT_ID>

Arguments:
  <CONNECTOR>  Owning connector ID from connector list/get
  <GRANT_ID>   Compute-grant ID from connector grant list/create, not a context ID

Options:
      --api <API>
          User API origin override; credentials never cross origins. Host/runner commands use explicit configuration
      --max-queued-runs <MAX_QUEUED_RUNS>
          Maximum accepted queued runs covered by this limit; lowering it preserves accepted work
      --max-outstanding-assignments <MAX_OUTSTANDING_ASSIGNMENTS>
          Dispatch capacity for unfinished assignments, including queued Slurm allocations
      --profile <PROFILE>
          Named local user API/login profile; host/runner commands use explicit configuration [default: default]
      --config <CONFIG>
          Private user profiles file; defaults to $XDG_CONFIG_HOME/enumeray/profiles.json or ~/.config/enumeray/profiles.json
      --generation <GENERATION>
          Observed compute-grant generation from connector grant get
      --json
          Emit user API results as JSON (NDJSON for following logs) and runtime errors on stderr; host/runner output follows its own protocol
      --timeout <TIMEOUT>
          User API HTTP timeout in seconds; live logs use it for read inactivity. Does not limit execution time [default: 60]
  -h, --help
          Print help