Test hang: cancelling_remote_dispatch_retires_the_borrowed_authenticated_channel hangs on dev (order-dependent) #333

New issue

Closed

opened 2026-11-03 21:11:31 +00:00 by kayg · 1 comment

kayg commented

2026-11-03 21:11:31 +00:00

Owner

daemon::remote_tests::cancelling_remote_dispatch_retires_the_borrowed_authenticated_channel (infraternal-agent --lib) hangs, and the hang depends on test order. Found by J152 while reviewing #281 (comment 77190). It is filed here as its own issue so the #281 branch doesn't carry it. J155 fixes it; J62d does not.

Evidence from J152, compute01, with INFRATERNAL_TEST_STATE_ROOT set and a 240 s per-binary watchdog:

Run alone (timeout 60): unmodified dev hung 3 of 3 times, and #281 head hung 3 of 3.
With daemon::remote_tests run as a module: dev 3 of 3, head 3 of 3.
In the full --lib run, dev's test order happens to avoid it (0 of 5). #281 head's slightly different test set doesn't (3 of 5, every one of those failing runs hung in this test). With only this test skipped, head passes all 65 remote_client tests.

So it is a pre-existing test-isolation problem, not branch code. #281 touches no daemon*/remote* code except one line in remote_supervisor_review_tests.rs. Once #281 merges, the gate gets flakier until this is fixed.

Wanted: a test-isolation fix, plus a bounded timeout on this test's awaits so a regression fails instead of hanging the gate.

`daemon::remote_tests::cancelling_remote_dispatch_retires_the_borrowed_authenticated_channel` (`infraternal-agent --lib`) hangs, and the hang depends on test order. Found by J152 while reviewing #281 (comment 77190). It is filed here as its own issue so the #281 branch doesn't carry it. **J155 fixes it**; J62d does not. Evidence from J152, compute01, with `INFRATERNAL_TEST_STATE_ROOT` set and a 240 s per-binary watchdog: - Run **alone** (`timeout 60`): unmodified dev hung 3 of 3 times, and #281 head hung 3 of 3. - With `daemon::remote_tests` run as a module: dev 3 of 3, head 3 of 3. - In the full `--lib` run, dev's test order happens to avoid it (0 of 5). #281 head's slightly different test set doesn't (3 of 5, every one of those failing runs hung in this test). With only this test skipped, head passes all 65 `remote_client` tests. So it is a pre-existing test-isolation problem, not branch code. #281 touches no `daemon*`/`remote*` code except one line in `remote_supervisor_review_tests.rs`. Once #281 merges, the gate gets flakier until this is fixed. Wanted: a test-isolation fix, plus a bounded timeout on this test's awaits so a regression fails instead of hanging the gate.

kayg referenced this issue

2026-11-03 21:43:49 +00:00

Reverse proxy: Traefik on demand — implementation plan #281

kayg referenced this issue

2026-11-03 22:51:55 +00:00

Reverse proxy: Traefik on demand — implementation plan #281

kayg commented

2026-11-04 17:04:55 +00:00

Author

Owner

J210 owner-authorized housekeeping: closing this duplicate record as superseded by #332, which tracks the same work and retains its implementation/evidence. This closure does not assert that all successor follow-ups are complete.

J210 owner-authorized housekeeping: closing this duplicate record as superseded by [#332](https://git.kayg.org/kayg/infraternal/issues/332), which tracks the same work and retains its implementation/evidence. This closure does not assert that all successor follow-ups are complete.

kayg closed this issue

2026-11-04 17:04:55 +00:00

kayg referenced this issue

2026-11-04 17:07:06 +00:00

J210: overnight housekeeping, 2026-11-04 #350

No labels

launch-qa

No milestone

No project

No assignees

1 participant

Notifications

Due date

The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set

Reference

kayg/infraternal#333

No description provided.

Rows
Columns