• Joined on 2026-03-08
ltms closed pull request fleet/fleetd#346 2026-09-04 11:14:53 +02:00
fleetd#341: per-name guard so a later spawns different unprotected credential name still warns
ltms closed pull request fleet/fleetd#344 2026-09-04 11:14:47 +02:00
fleetd #339: guard backend error sink
ltms closed issue fleet/fleetd#341 2026-09-04 11:14:41 +02:00
One flag guards two different memberCredentials WARNs, so a later leaking credential name is never reported
ltms closed issue fleet/fleetd#339 2026-09-04 11:14:36 +02:00
A member's normal report that quotes "API Error:" is recorded as a real backend outage
ltms commented on issue fleet/fleetd#341 2026-09-04 11:14:30 +02:00
One flag guards two different memberCredentials WARNs, so a later leaking credential name is never reported

Merged to main as b32a30f (--no-ff; the branch was behind main). Follow-up commit d057d56 adds two tests — see below.

Build: Tests run: 1362, Failures: 0, Errors: 0, Skipped: 0, `BUILD…

ltms commented on issue fleet/fleetd#339 2026-09-04 11:14:11 +02:00
A member's normal report that quotes "API Error:" is recorded as a real backend outage

Merged to main as d11d1d1 (--no-ff; the branch was behind main). Follow-up commit 6a81417 corrects a regression the fix introduced — read that part.

Build after both: `Tests run:…

ltms opened issue fleet/fleetd#348 2026-09-04 11:13:50 +02:00
The same text-match-drives-a-cooldown shape exists for ExhaustionSink, where the penalty is 30x longer
ltms pushed to main at fleet/fleetd 2026-09-04 11:13:24 +02:00
6a814176f0 #339: the start-of-line check must skip terminal chrome
d11d1d157c Merge #339: a backend-error text match must be at the start of its line before it records a credential outage
d057d56156 #341: pin the noise control, and the reverse policy order
b32a30fd47 Merge #341: warn once per distinct unprotected credential name, not once per launcher
464dbc0930 fleetd#341: a per-name guard so a later spawn's different unprotected name still warns
Compare 6 commits »
ltms closed pull request fleet/fleetd#343 2026-09-04 11:01:33 +02:00
#338: cancel timed-out queued deliveries
ltms closed issue fleet/fleetd#338 2026-09-04 11:01:27 +02:00
A TIMED_OUT_QUEUED send leaves its message in the injector queue, and it is delivered later as a stale brief
ltms commented on issue fleet/fleetd#338 2026-09-04 11:01:22 +02:00
A TIMED_OUT_QUEUED send leaves its message in the injector queue, and it is delivered later as a stale brief

Merged to main as a8cadd9. The branch was up to date with main — checked with git merge-base --is-ancestor, not assumed.

Build after the merge, unpiped: `Tests run: 1357, Failures: 0,…

ltms opened issue fleet/fleetd#345 2026-09-04 11:01:04 +02:00
MessageService's use of Cancellation.DELIVERED is unpinned — the seam is tested, the caller is not
ltms pushed to main at fleet/fleetd 2026-09-04 11:00:39 +02:00
a8cadd9150 Merge #338: a timed-out queued send cancels its message instead of leaving it to be delivered later
147f50c19e #338: cancel timed-out queued deliveries
Compare 2 commits »
ltms opened issue fleet/fleetd#342 2026-09-04 10:51:36 +02:00
CompositePeerLauncher's single-daemon stop() shortcut can skip tab teardown, orphaning a worker tab
ltms opened issue fleet/fleetd#341 2026-09-04 10:51:10 +02:00
One flag guards two different memberCredentials WARNs, so a later leaking credential name is never reported
ltms opened issue fleet/fleetd#340 2026-09-04 10:48:45 +02:00
Two ways the injector can type into a member that is mid-turn — measure before fixing
ltms opened issue fleet/fleetd#339 2026-09-04 10:48:09 +02:00
A member's normal report that quotes "API Error:" is recorded as a real backend outage
ltms opened issue fleet/fleetd#338 2026-09-04 10:47:42 +02:00
A TIMED_OUT_QUEUED send leaves its message in the injector queue, and it is delivered later as a stale brief
ltms closed pull request fleet/fleetd#336 2026-09-04 10:44:41 +02:00
fleetd#333: fleet.leaders is split, and split membership now proves reporting exists
ltms closed issue fleet/fleetd#333 2026-09-04 10:44:36 +02:00
fleet: is a third split key, sitting in the coverage checker's own escape hatch — and SPLIT_KEYS membership does not prove any reporting code exists