agent-substrate / agent-substrate/substrate

TestActorSnapshotLifecycle reads only the first page of ListActorSnapshots

Open
#1,126 0 comments 0 reactions 0 assignees View on GitHub
area/storage area/tests kind/bug
Dominant language
Go
Stars
1.8k
Forks
316
Avg merge
2d 43m
Merged PRs (30d)
287

Description

`TestActorSnapshotLifecycle` fails with `snapshot %q missing from ListActorSnapshots` on a cluster that has been up a while.

The snapshot is there. `GetActorSnapshot` succeeds on it one line earlier. The test then calls `ListActorSnapshots` once and treats the first page as the whole list, but the list is name-ordered and the API says there is more in `NextPageToken`. On the cluster where this reproduced, the newly created snapshot sorted 1408th of 1518 against a page size of 1000 — page two.

`createActor` reads `ListActors` the same way.

CI never sees it: every run gets a fresh cluster, and the `demo` atespace stays far below one page. It bites long-lived and shared clusters, and it cost time during the #1050 investigation by looking like a real snapshot loss.

Measured on the cluster where it reproduced: 15/35 runs failed before, 0/35 after following the page tokens (Fisher one-sided p = 4.5e-6).

Contributor guide

Open the contributing guide

Assessment

This issue has not been assessed yet.

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.