Manage compute instances¶
Compute instances provide runtime capacity for SQL, workflows, and data-processing tasks. Read the available specification and Worker-image lists before creating an instance; ordinary callers select a platform-provided spec_id and do not create specifications.
Read specifications and Worker images
→ Create a compute instance and save resource_id
→ Query details, runtime state, or metrics
→ Suspend, resume, or retry when appropriate
→ Run deletion preflight
→ Delete the instance
Creation and lifecycle requests can finish asynchronously. A successful HTTP response means that the service accepted the request, not that the instance is ready. Continue reading status and status_message on the instance detail.
Before deletion, call the delete preflight endpoint and resolve active-task and workflow dependencies. Deletion is irreversible.
Last updated on