ActiveJobKey |
Identifies a specific instance of a Job. (jobId, role) is unique within Manager; a 2-replica bundle has two ActiveJobKeys sharing jobId but with different role (e.g. "primary"/"backup"). During a Phase E migration the source keeps its role and the target wears role="migrationTarget". |
AuditAppendedEvent |
A new audit-log entry was appended on the daemon (any source). Unlike the AutoManager raw-PB events, this carries the SDK’s own AuditLogEntry — the audit domain type lives in the SDK (it’s what getAuditLog returns), so the conversion happens here rather than in consumers. |
AuditLogEntry |
/** One entry in the daemon’s durable, hash-chained audit log. seq is narrowed from the wire int64 to a number (entry counts stay well within 2^53). payload is opaque (already-serialised JSON or text). |
AwsHealth |
|
AwsLaunchTemplateId |
|
AwsLaunchTemplateName |
|
AwsNodeCreate |
|
BlueprintRef |
Identifies a blueprint tar — opaque to Manager, resolved by the worker at launch time. sha256 is the canonical cache key (lowercase hex of the tar’s SHA-256). The same tar may arrive via different transports (URL, inline bytes, content-addressed lookup) but its cache entry is one and the same.
See .ai/steve/tasks/39-blueprint-jobs.md for the full design. |
BundleDeletedEvent |
|
BundleUpdatedEvent |
|
ClusterNodeInfo |
Snapshot of a cluster slot returned from listClusterNodes.
active is true when the worker’s gRPC stream is currently connected. When inactive, listenAddress / listenPort / hardware reflect the last known values (or are empty/zero if the worker has never registered).
|
ClusterRegistration |
|
CurrentJob |
|
DateFilter |
|
EventStreamClosed |
The manager activity stream has closed |
EventStreamInitialState |
|
EventStreamSettings |
Configuration for creating an event stream |
GpuInfo |
Hardware capabilities a cluster worker reports during registration. v1 carries cpu count + total memory; gpu / sdi / network arrays are placeholder shapes for a later sub-phase. |
HardwareInfo |
|
HydrationDrop |
A persisted row a daemon store couldn’t parse on the current boot (schema drift / corruption). The live set — clears when a later boot hydrates cleanly. key is the RocksDB key; store the producer. |
IdFilter |
|
Job |
|
JobConfigUpdatedEvent |
|
JobDeleted |
A job has been deleted |
JobInfo |
Informational messages about a job |
JobOutOfWindow |
A job has been updated such that it no longer matches the search criteria |
JobPending |
Job is almost ready to start - "almost" defined by property passed into the list functions - todo, sort out this description |
JobRejectedEvent |
|
JobSearchSettings |
|
JobUpdated |
Job details have been updated |
JobWithHistory |
|
MigrationAbortedEvent |
Phase E migration event. A worker workflow has reported it cannot continue the migration handshake; AutoManager aborts. |
MigrationSourceStoppedOutputEvent |
Phase E migration event. Worker has signalled that the source instance stopped output at last_output_timestamp_ns. AutoManager forwards that timestamp to the target via NotifyMigrationCutoverApply. |
MigrationTargetReadyEvent |
Phase E migration event. Worker has signalled that a role="migrationTarget" instance is ready to begin output. AutoManager owns the protocol; SDK consumers see the raw PB. |
NetworkInfo |
|
NodeInventoryUpdatedEvent |
|
NodePending |
A node is pending |
NodeRunning |
A node is running - used in initial state messages |
NodeServiceDeletedEvent |
|
NodeServiceUpdatedEvent |
|
NodeStarting |
A node is starting |
NodeStopped |
A node has stopped |
NodeStopping |
A node is stopping |
NoFilter |
|
NorskSettings |
Top level Norsk configuration |
OciHealth |
|
OciNodeCreate |
|
PhysicalNodeConnected |
A physical node has connected |
ProviderHealthChange |
;
The health status of a provider has changed |
SdiInfo |
|
StateFilter |
|
TagFilter |
|
VersionInfo |
Norsk Manager Version information |