ltms - Gitea

ltms commented on pull request fleet/fleetd#200

2026-08-31 17:07:29 +02:00

#164: classify a backend-error scrape as WORKER_FAILED, not a completed reply

Already on main — closing as superseded, not rejected. The change landed in 4ac688b ("#164: classify a backend-error scrape as WORKER_FAILED, carrying the whole pane"), byte-identical to this…

ltms closed pull request fleet/fleetd#204

2026-08-31 17:06:53 +02:00

CB-172: block broker URI env from members

ltms commented on pull request fleet/fleetd#204

2026-08-31 17:06:46 +02:00

CB-172: block broker URI env from members

Already on main — closing as superseded, not rejected.

Every line of this PR's substance landed in ad587ea ("#172: keep the broker URI, password and all, out of every member pane"):…

ltms commented on issue fleet/fleetd#175

2026-08-31 17:01:07 +02:00

fleetd asks opencode for a model and never checks it got that model, so a withdrawn name silently bills a paid credential

Status update. PR #203 is closed as superseded — see the comment there for the detail. Two things changed under it:

Its read-back ran inside spawn(), before opencode writes the session row,…

ltms closed pull request fleet/fleetd#203

2026-08-31 17:00:46 +02:00

CB-175: quarantine opencode model mismatches

ltms commented on pull request fleet/fleetd#203

2026-08-31 17:00:34 +02:00

CB-175: quarantine opencode model mismatches

Closing this as superseded. Two independent reasons, neither of which is a criticism of the work — the code reads correctly and the tests are real.

**1. The guard can never fire, by construction.…

ltms closed issue fleet/fleetd#206

2026-08-31 16:59:36 +02:00

opencode session discovery never finds a member's record, so agentSessionId is always null and resume silently does nothing

ltms commented on issue fleet/fleetd#206

2026-08-31 16:59:30 +02:00

opencode session discovery never finds a member's record, so agentSessionId is always null and resume silently does nothing

Fixed by PR #207, merged to main in a55079a, daemon redeployed onto jar 9ec5fab0f136.

The discovery itself is proven live: an opencode member spawned on a fresh worktree produced the…

ltms closed pull request fleet/fleetd#208

2026-08-31 16:58:54 +02:00

fleetd #185 stage 3: opt-in worktreeGroup for group-shared worktrees

ltms commented on pull request fleet/fleetd#208

2026-08-31 16:58:48 +02:00

fleetd #185 stage 3: opt-in worktreeGroup for group-shared worktrees

Merged to main manually, integrated with #207 into a55079a. Full suite on main: 1061 tests, 0 failures. Daemon redeployed onto that jar (9ec5fab0f136).

Two defects were fixed on top of this…

ltms closed pull request fleet/fleetd#207

2026-08-31 16:58:34 +02:00

fleetd #206: read opencode sessions from opencode.db instead of the frozen JSON tree

ltms commented on pull request fleet/fleetd#207

2026-08-31 16:58:27 +02:00

fleetd #206: read opencode sessions from opencode.db instead of the frozen JSON tree

Merged to main manually as e18ad47 (head a052975), integrated with #208 into a55079a. Full suite on main: 1061 tests, 0 failures. Daemon redeployed onto that jar (9ec5fab0f136).

Two…

ltms opened issue fleet/fleetd#209

2026-08-31 16:57:12 +02:00

agentSessionId is resolved once at spawn and frozen, so an opencode member never reports one

ltms pushed to main at fleet/fleetd

2026-08-31 16:49:39 +02:00

a55079afbd #185: opt-in worktreeGroup, so a member running as another OS user can write its worktree

e18ad4723b Merge remote-tracking branch 'refs/remotes/origin/cb206'

6de8ac8972 #206: read opencode session ids from opencode.db, and pin the read-only open

a052975420 #206: pin the read-only open with a test that actually fails without it

6d82ca95a4 #185: skip absent git paths, and resolve the git dir instead of assuming .git

Compare 7 commits »

ltms pushed to worker/cb206-opencode-sqlite-128718-2 at fleet/fleetd

2026-08-31 16:46:15 +02:00

a052975420 #206: pin the read-only open with a test that actually fails without it

ltms pushed to worker/cb185-worktree-group-fc0c99-1 at fleet/fleetd

2026-08-31 16:35:04 +02:00

6d82ca95a4 #185: skip absent git paths, and resolve the git dir instead of assuming .git

ltms commented on issue fleet/fleetd#185

2026-08-31 16:31:50 +02:00

Run members on a second herdr under a dedicated user (optional; single-herdr stays the default)

Stage 3 in progress, and four more files that assume fleetd's uid owns everything it writes

Stage 3 (worktree group/setgid provisioning) is implemented in PR #208: an opt-in worktreeGroup: ke…

ltms commented on issue fleet/fleetd#206

2026-08-31 10:44:11 +02:00

opencode session discovery never finds a member's record, so agentSessionId is always null and resume silently does nothing

Root cause found: opencode moved its session store from JSON files to SQLite

Candidate 3 was the right one — "the layout moved". Candidates 1 and 2 are ruled out.

**opencode now keeps…

ltms commented on issue fleet/fleetd#161

2026-08-31 09:34:04 +02:00

a pane member's grandchild process may be resolved as primary — possible worker→primary escalation

Confirmed live on the running daemon — the half this ticket could never prove

The fix is deployed (pid 19293, jar 509df0608da2, 1049 tests) and I have now run the end-to-end probe this…

ltms commented on issue fleet/fleetd#137

2026-08-31 09:28:14 +02:00

CB-629: after a bridge_ask round-trip, the worker's final reply orphans its ticket, which then reports a false failure

Fixed and merged to main as 388aba7. Closing.

The hypothesis in this ticket was close, but the real cause is narrower

This ticket guessed that the fleet_ask reverse rendezvous detaches…