Organization API: Organization deployments: Logs

Part of Organization deployments.

GET /api/v1/orgs/{org_slug}/deployments/{deployment_uuid}/logs

The newest limit lines that match, oldest first within the page so a terminal prints them in order, and next: the cursor for the page before it (pass it as before), or null at the oldest line kept. Lines are kept for the plan's deploy_log_retention_days (retention_days); partial says a stored chunk could not be read or the page stopped at its read budget, and next continues after it. Each line: id (<chunk>:<n>), ts (when it was collected, within a minute of when it was printed), instance, process, location, release, message, and gap (lines may be missing just before it). As NDJSON the lines are the body and next is the X-Coritan-Next header. 503 logs_unavailable when the log bucket is not set up.

GET /api/v1/orgs/{org_slug}/deployments/{deployment_uuid}/logs/instances

The deployment's instances that are live or have lines kept, newest first: id (what instance takes), location, process, release, state, live, sleeping, created_at and last_line_at.

GET /api/v1/orgs/{org_slug}/deployments/{deployment_uuid}/logs/stream

text/event-stream of the running instances' output (ten at most), polled from their consoles every two seconds: ready (instances, until, poll_seconds), then line events shaped as /logs lines (id is live:<instance>:<n>), instance when one's console cannot be read, a : keepalive comment after fifteen quiet seconds, and end with reason: time_limit after seconds, gone when the deployment is no longer the caller's, unavailable when consoles cannot be read here. Reconnect for more; a tail does not replay what an earlier one sent. 429 too_many_streams past five open at once for the account (on an organization's routes, for the organization), whichever keys, sign-ins or members opened them, and rate_limited past thirty opened by one caller in ten minutes.