Repository navigation
fix: ADR 0044 K5 follow-ups — operator /api/v1 paths + eviction-lease state re-check - #128
Merged
Merged
Conversation
The K3 operator's coord client built paths like /api/admin/hosts/:id/cordon and /api/hosts/:id — missing the /api/v1 prefix the whole coordinator router is nested under (api/mod.rs). The dev-vm mock coord returned 200 for any path and masked it; against the real coord every call 404s, so the operator's drain-gated rollout (and the K4 fleet-demand poll) silently fail. Caught by the live K5 cutover. Centralise the prefix in a `url()` helper + a `urls_are_v1_prefixed` test so a path typo can't slip past a permissive mock again. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
… K5) evict_session_to_state had a stale-binding guard on the in-memory registry but not on the authoritative PG state. When an idle-evict completed exactly as an admin drain dispatched the same session, the drain acquired the (just-released) lease and drove an already-Idle session to Evacuating on a destroyed sandbox — wedging it (Evacuating, no sandbox, no snapshot). The lease is held until after the PG transition, so once held, any concurrent eviction has fully completed. Re-read the session and skip unless it's still Active or Evicting (the two legal inputs: direct idle-evict/drain, and backstop-nominated). idle-evict and drain now compose toward running_sandboxes=0 instead of colliding. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Two bugs the live K5 cutover surfaced:
/api/v1— the K3 operator's coord client called/api/admin/...//api/hosts/:idwithout the version prefix; the dev-vm mock coord (accepts any path) masked it, but it 404s against the real coord, so drain-gated rollout + the K4 demand poll fail. Fixed via a centralizedurl()helper + a path-assertion test.evict_session_to_stateonly guarded the in-memory binding, not the PG state; an idle-evict finishing as a drain dispatched the same session wedged it inEvacuatingwith no sandbox. Now re-reads the session under the lease and skips unlessActive | Evicting.just checkgreen (1034 tests, incl. the new operator path test). Merging rebakes host-operator + coordinator and auto-rolls the prod coord with fix #2.🤖 Generated with Claude Code