ltms - Gitea

ltms closed issue fleet/fleetd#297

2026-09-04 07:20:14 +02:00

REST is the weaker door, and it is the door a lead falls back to when MCP drops

ltms commented on issue fleet/fleetd#297

2026-09-04 07:19:32 +02:00

REST is the weaker door, and it is the door a lead falls back to when MCP drops

Merged as f34361b, corrected in 9debc0d. Real merge built green at 1307 tests.

The wiring is right, and it is the part I checked hardest. Fleetd now names quarantineSource and…

ltms pushed to main at fleet/fleetd

2026-09-04 07:19:07 +02:00

9debc0de27 #297: render both profiles doors from one body builder, not two copies

f34361b263 Merge #297: REST reports what MCP reports

85c90d440a #297: map HerdrException on GET /agents and /members; GET /profiles reports quarantine + cool-off

Compare 3 commits »

ltms closed pull request fleet/fleetd#299

2026-09-04 07:10:19 +02:00

fleetd #296: close panes on failed spawn

ltms closed issue fleet/fleetd#296

2026-09-04 07:10:16 +02:00

Two spawn-side exits leak a live pane, and the caller cannot clean up because it never learns the pane id

ltms commented on issue fleet/fleetd#296

2026-09-04 07:10:09 +02:00

Two spawn-side exits leak a live pane, and the caller cannot clean up because it never learns the pane id

Merged as aef14ff. Real merge built green at 1304 tests.

Both exits now close the pane they opened, best-effort, catching RuntimeException, logging a WARN that names the pane, and…

ltms pushed to main at fleet/fleetd

2026-09-04 07:09:49 +02:00

e97502d550 drainAll: record that the timeout is a whole-drain budget, not a per-session grace

aef14ff46e Merge #296: a failed spawn no longer leaks the pane it opened

cba516bda4 fleetd #296: close panes on failed spawn

Compare 3 commits »

ltms opened issue fleet/fleetd#298

2026-09-04 07:05:57 +02:00

AmqpReplyInbox.release() drops an unacked reply that the broker still holds — a worker's report is lost silently

ltms opened issue fleet/fleetd#297

2026-09-04 07:02:57 +02:00

REST is the weaker door, and it is the door a lead falls back to when MCP drops

ltms opened issue fleet/fleetd#296

2026-09-04 07:00:14 +02:00

Two spawn-side exits leak a live pane, and the caller cannot clean up because it never learns the pane id

ltms closed pull request fleet/fleetd#295

2026-09-04 06:50:47 +02:00

#293: wrap the bare tab.close in HerdrPeerLauncher.stop()

ltms closed issue fleet/fleetd#293

2026-09-04 06:50:45 +02:00

A failing tab.close during teardown leaks both the ZDOTDIR and the worktree

ltms commented on issue fleet/fleetd#293

2026-09-04 06:48:11 +02:00

A failing tab.close during teardown leaks both the ZDOTDIR and the worktree

Merged as 086c598, extended in ba51e0c. Real merge built green at 1303 tests.

The worker read every caller I named before deciding, rather than applying the obvious fix and moving on. Its…

ltms pushed to main at fleet/fleetd

2026-09-04 06:47:46 +02:00

ba51e0c6cc #293: catch RuntimeException on closeTab, matching its sibling guard

086c59848e Merge #293: a failing tab.close no longer masks the cleanups below it

0c10079755 #293: wrap the bare tab.close in HerdrPeerLauncher.stop()

Compare 3 commits »

ltms closed pull request fleet/fleetd#294

2026-09-04 06:33:51 +02:00

#280: sweep a GONE/NEVER_READY target's lapsed fleet_ask once, not just on transition

ltms closed issue fleet/fleetd#280

2026-09-04 06:33:48 +02:00

Follow-up to #275: a health-detected GONE member whose ask lapses may still strand its ticket

ltms commented on issue fleet/fleetd#280

2026-09-04 06:33:36 +02:00

Follow-up to #275: a health-detected GONE member whose ask lapses may still strand its ticket

Merged as b3f917e, corrected in ece2091. Real merge built green at 1301 tests.

Step 1 was answered, and the answer is "reachable"

This was the whole ticket, and the worker did the…

ltms pushed to main at fleet/fleetd

2026-09-04 06:33:01 +02:00

ece2091b53 #280: states is now touched by two scheduler tasks, so make it concurrent

b3f917e6f5 Merge #280: one bounded delayed re-check sweeps a ticket whose ask lapsed after GONE

fa39a5f55e #280: sweep a GONE/NEVER_READY target's lapsed fleet_ask once, not just on transition

Compare 3 commits »

ltms closed pull request fleet/fleetd#292

2026-09-04 06:28:32 +02:00

#290: restore reapIdle per-session guard coverage via a launcher.stop() trigger

ltms closed issue fleet/fleetd#290

2026-09-04 06:28:29 +02:00

Restore a test for reapIdle's per-session guard, which #283 left uncovered