Actions
Cedana API for actions
Only return actions that ran on this cluster
Returns the actions with status
The checkpoint's CEDANA_CHECKPOINT name (the env value it restores from), used to group automatic/named checkpoints. Null for unnamed checkpoints.
Whether the checkpoint is a (GPU delta) increment. Authoritative even when parent_checkpoint_id is null (malformed id at ingest, parent deleted)
Checkpoint this one is an increment of; null for full checkpoints and for deltas whose parent is unknown or deleted
Owning workload kind of the checkpointed pod: "DynamoGraphDeployment" when the pod belongs to a Dynamo deployment, else the pod's immediate ownerReferences kind (ReplicaSet/Job/StatefulSet…). Null if unknown.
Failed list
Failed to connect/fetch from jobs table from database
Client error
Server error
GET /v1/actions HTTP/1.1
Authorization: Bearer YOUR_SECRET_TOKEN
Accept: */*
<uuid-v4-here>The action_id of the given pod_id has been fetched successfully
Failed to get action_id for the given pod_id
Failed to connect/fetch from pods_to_actions table from database
Client error
Server error
GET /v1/actions/from_pod/{id} HTTP/1.1
Authorization: Bearer YOUR_SECRET_TOKEN
Accept: */*
<uuid-v4-here>Success
Failed to get action_id within the db
Failed to connect/fetch from pods_to_actions table from database
Client error
Server error
POST /v1/actions/upload HTTP/1.1
Authorization: Bearer YOUR_SECRET_TOKEN
Content-Type: application/json
Accept: */*
Content-Length: 122
{
"action_id": "text",
"persistent_volume_claim": null,
"pod_id": null,
"pod_metadata_spec": null,
"pod_spec": null,
"status": "text"
}SuccessReturns detailed information about a specific action including associated checkpoint data
Action UUID
Action details retrieved successfully
Action details response for a single action
The action ID
When the action was created
When checkpoint completed (if any)
Associated checkpoint ID (if any)
Full checkpoint profiling info (JSON)
Checkpoint name/image
Checkpoint size in bytes
Original request details (JSONB)
GPU type used
Node where the pod was running
Platform (e.g., containerd)
Reason for the action (heartbeat, manual, nodeTermination, etc.)
Restore path for the checkpoint
Current status of the action
Checkpoint duration in nanoseconds (from profiling)
Total I/O in bytes (from profiling)
Action type (checkpoint_pod, restore_pod, heartbeat)
Invalid action ID format
Action not found
Database error
Client error
Server error
GET /v1/actions/{id} HTTP/1.1
Authorization: Bearer YOUR_SECRET_TOKEN
Accept: */*
{
"action_id": "123e4567-e89b-12d3-a456-426614174000",
"action_timestamp": null,
"checkpoint_completed_timestamp": null,
"checkpoint_id": null,
"checkpoint_info": null,
"checkpoint_name": null,
"checkpoint_size": null,
"details": null,
"gpu": null,
"node_name": null,
"platform": null,
"reason": null,
"restore_path": null,
"status": "text",
"total_duration": null,
"total_io": null,
"type": "text"
}Returns the path of the latest successful checkpoint for a given checkpoint name
Returns the restore path
No ready checkpoint with the given name
Failed to connect/fetch from jobs table from database
Client error
Server error
GET /v1/checkpoint/path HTTP/1.1
Authorization: Bearer YOUR_SECRET_TOKEN
Accept: */*
<restore-path>Returns the ID for the task
Invalid request parameters
Failed to find checkpointable resource
Failed to create checkpoint or connect to eventstream
Client error
Server error
POST /v1/checkpoint/pod HTTP/1.1
Authorization: Bearer YOUR_SECRET_TOKEN
Content-Type: application/json
Accept: */*
Content-Length: 120
{
"cluster_id": null,
"kind": "simple",
"namespace": null,
"overrides": null,
"pod_id": null,
"pod_name": null,
"reason": "heartbeat"
}550e8400-e29b-41d4-a716-446655440000Returns the current status of a checkpoint action
Returns the current status of the checkpoint operation
Action ID not found
Failed to retrieve status from database
Client error
Server error
GET /v1/checkpoint/status/{action_id} HTTP/1.1
Authorization: Bearer YOUR_SECRET_TOKEN
Accept: */*
{
"details": null,
"status": "initialized"
}Returns the action ID for the new restore
Invalid action_id or request parameters
Failed to create restore or connect to eventstream
Client error
Server error
POST /v1/restore/pod HTTP/1.1
Authorization: Bearer YOUR_SECRET_TOKEN
Content-Type: application/json
Accept: */*
Content-Length: 84
{
"action_id": "text",
"cluster_id": "text",
"overrides": null,
"reason": "nodeTermination"
}550e8400-e29b-41d4-a716-446655440000Last updated
Was this helpful?