Fleet queue entry

Included from Basic tier

One dispatch-plane entry as GET /api/fleet/queue (admin token) serves it: a step the scheduler currently carries — parked ready or held by a live lease — annotated with its placement verdict. This page is the field dictionary only; the narrative (where entries come from, the verdict ladder, the levers, failure modes) lives at opschain docs fleet-queue and is not repeated here.

The v2 OpenAPI document declares this resource's schema; the resource-field coverage gate holds this page to it.

The response envelope is {"entries": [...], "halted": ..., "quiesced": ..., "quiesced_by": ..., "quiesced_at": ...} — see opschain docs fleet-queue for the envelope fields.

Entry fields

Field	Type	Meaning	Origin / null semantics
run_id	string	The owning run (change or workflow run).	From the queue payload written at enqueue.
step_id	string	The step this entry dispatches.	Same.
attempt	int	The attempt number — a retried step re-enters as a new attempt.	Same.
queue	string	The scheduler's internal dispatch-queue name (default today; reserved for future named queues).	The queue_ready / scheduler_job_leases row.
ready_at	timestamp	When the entry became (or becomes) eligible — backoff retries park with a future value.	Queue payload. Always present.
leased	bool	false: parked ready in queue_ready. true: a runner holds the job lease (scheduler_job_leases).	Which table the entry was read from.
requirement	object	The step's placement demand, resolved once at enqueue (later settings edits never reclassify queued work). Serialized {} when it demands nothing. Sub-fields below.	Queue payload.
satisfiable	bool	Whether anything can take the entry right now. Always true on leased entries; on ready entries, true exactly when reason is empty.	Computed per read.
reason	string	Why the entry is not dispatching — the verdict ladder's first match (halt, system quiesce, licence stop, replica quiesce, the pinned-runner verdict, then the placement verdicts). Absent (omitted) when satisfiable, and always absent on leased entries.	Computed per read; strings quoted verbatim in opschain docs fleet-queue.
pinned_runner	string	The runner this entry's run is pinned to (run-scoped execution); the entry is only ever offered to it. Omitted when unpinned.	The scheduler's scheduler_run_pins row, joined per read.
run_scoped	bool	Whether the entry's run is run-scoped — one runner, one shared workspace: the run pins to the first runner that takes one of its steps. Omitted when false.	Queue payload, classified at enqueue.
scope_source	string	Why the run's execution scope resolved as it did: runner.execution_scope override when the setting decided it, else the derivation's deciding fact — derived: actions.rb-backed tree (discovered/catalog change), derived: ruby-lane steps present (a declared tree the creation census found an actions.rb#/mintpress# step in), derived: declarative tree, or derived: workflow runs are never pinned. The actions.rb/ruby-lane values are only applicable to existing MintPress customers and those using the MintPress SDK or the packs:mintpress feature. Omitted when the owning run could not be read at enqueue.	Queue payload, classified at enqueue.
owner_type	string	The owning work item's kind: change, workflow_run or node_background_task (a task's synthetic job carries the task id as run_id). The owner's id IS run_id.	Resolved per read across the three owning tables in one indexed statement. Omitted — with every other owner field — when run_id matches no owning row this store holds.
owner_status	string	The owner's current status: a run's engine status (pending/queued/running/waiting/…), or the task's status_code.	Same resolution. Omitted when the owner is unresolved.
owner_terminal	bool	Whether the owner is already finalised. A parked entry with a finalised owner is an orphan — SweepQueueOrphans / opschain admin queue sweep clears it.	Classified against the owner kind's live set (pending/queued/running/waiting for runs, initializing/running for tasks). Omitted when false or unresolved.
target_path	string	The owner's target node path.	The change's target node / the run record's target path / the task's asset node. Omitted when the owner is unresolved.
correlation_id	string	The owner's correlation chain id, computed read-time as coalesce(correlation_id, id) for runs.	Omitted for node background tasks (no chain) and unresolved owners.

The scheduler's internal QueueEntry also carries a priority; it is not serialized into this payload.

requirement sub-fields (all omitted when empty)

Field	Type	Demand
group	string	The runner's group must equal this name.
groups	string array	The runner's group must be one of this set (the node-path-binding shape).
labels	string array	Every listed label must be present on the runner.
executors	string array	Every listed executor kind must be advertised by the runner (the step's own kind is folded in at enqueue).
selectors	map of string → string array	Group-attribute demands: each key must resolve to a runner-group attribute whose value is in the key's list (equality and in-set only).
secrets_providers	string array	Every listed secrets provider code must be declared by the runner: the step's secret references route to these runner-resolved providers, resolved once at enqueue from the run's property documents.

Related pages

• opschain docs fleet-queue — the full narrative: enqueue, the verdict ladder with every reason string verbatim, requirement resolution order, levers and failure modes, and the termination / orphan-sweep verbs.
• opschain docs dispatch — the scheduler behind the queue.
• opschain docs cli/admin-queue-list — the same entries (typed, via the API v2 FleetQueueService) with the owner columns rendered.