fleet/fleetd
3
0
Fork 0
Code Issues 73 Pull Requests Actions Packages Projects Releases 1 Wiki Activity

fleetd #284: free backend-error capacity #287

Closed
agent wants to merge 0 commits from worker/fix-284-backend-error-seat-85912c-11 into main
pull from: worker/fix-284-backend-error-seat-85912c-11
merge into: fleet:main
Conversation 1 Commits - Files Changed -
agent commented 2026-09-04 05:47:47 +02:00
Member
Copy Link
Copy Source

Report

PR: #287
Branch: worker/fix-284-backend-error-seat-85912c-11
Root: /Users/dai.ha/LTMS/.bridged-worktrees/04928f-11
Files: fleetd/src/main/java/dev/ltms/fleet/Fleetd.java, fleetd/src/main/java/dev/ltms/fleet/mcp/FleetMcp.java, fleetd/src/test/java/dev/ltms/fleet/FleetdBackendErrorSinkTest.java, and fleetd/src/test/java/dev/ltms/fleet/mcp/FleetMcpTest.java.

Decision

I chose option 2. BACKEND_ERROR stays in the roster so fleet_list keeps its failure details. It no longer uses profile capacity. This avoids a teardown path that could remove a pane or worktree after a backend failure.

Change

  • Fleetd now excludes BACKEND_ERROR from the live counter used by the real CompositePeerLauncher maxLoad gate.
  • fleet_list reports BACKEND_ERROR sessions as reclaimable.
  • The gate-level test acquires a maxLoad-limited member, calls onBackendError, then acquires another member on the same profile. The second spawn succeeds.

Regression proof

I removed the BACKEND_ERROR exclusion temporarily and ran the focused test. It failed with:

worker profile 'terra' is at maxLoad: 1 live >= 1 cap; refusing spawn — no fallback to another profile

I restored the exclusion before the full build.

Checks

  • cd fleetd && mvn clean install
  • Tests run: 1285, Failures: 0, Errors: 0, Skipped: 0
  • BUILD SUCCESS

Caveat

FAILED is also set by SessionManager.onFailed, is non-deliverable, and is not reaped. It remains outside this issue scope.

## Report PR: https://git.ltms.dev/fleet/fleetd/pulls/287 Branch: `worker/fix-284-backend-error-seat-85912c-11` Root: `/Users/dai.ha/LTMS/.bridged-worktrees/04928f-11` Files: `fleetd/src/main/java/dev/ltms/fleet/Fleetd.java`, `fleetd/src/main/java/dev/ltms/fleet/mcp/FleetMcp.java`, `fleetd/src/test/java/dev/ltms/fleet/FleetdBackendErrorSinkTest.java`, and `fleetd/src/test/java/dev/ltms/fleet/mcp/FleetMcpTest.java`. ### Decision I chose option 2. `BACKEND_ERROR` stays in the roster so `fleet_list` keeps its failure details. It no longer uses profile capacity. This avoids a teardown path that could remove a pane or worktree after a backend failure. ### Change - `Fleetd` now excludes `BACKEND_ERROR` from the live counter used by the real `CompositePeerLauncher` maxLoad gate. - `fleet_list` reports `BACKEND_ERROR` sessions as reclaimable. - The gate-level test acquires a maxLoad-limited member, calls `onBackendError`, then acquires another member on the same profile. The second spawn succeeds. ### Regression proof I removed the `BACKEND_ERROR` exclusion temporarily and ran the focused test. It failed with: `worker profile 'terra' is at maxLoad: 1 live >= 1 cap; refusing spawn — no fallback to another profile` I restored the exclusion before the full build. ### Checks - `cd fleetd && mvn clean install` - `Tests run: 1285, Failures: 0, Errors: 0, Skipped: 0` - `BUILD SUCCESS` ### Caveat `FAILED` is also set by `SessionManager.onFailed`, is non-deliverable, and is not reaped. It remains outside this issue scope.
agent added 1 commit 2026-09-04 05:47:47 +02:00
fleetd #284: free backend-error capacity
CI / contract (pull_request) Successful in 46s
Details
CI / build (pull_request) Successful in 2m42s
Details
4dd12083ab
agent added 1 commit 2026-09-04 05:55:24 +02:00
fleetd #284: free failed-session capacity
CI / contract (pull_request) Successful in 1m11s
Details
CI / build (pull_request) Successful in 1m19s
Details
11cbfa79b4
ltms commented 2026-09-04 06:12:43 +02:00
Owner
Copy Link
Copy Source

Merged locally. Posting this here because the worker had no gh and could not update its own PR body — the round-two commit 11cbfa7 extends the fix from BACKEND_ERROR to FAILED as well, and the body still describes only the first round.

I also changed one thing the worker did on my own instruction, because the instruction was wrong. Reasoning is on #284.

Merged locally. Posting this here because the worker had no `gh` and could not update its own PR body — the round-two commit `11cbfa7` extends the fix from `BACKEND_ERROR` to `FAILED` as well, and the body still describes only the first round. I also changed one thing the worker did on my own instruction, because the instruction was wrong. Reasoning is on #284.
ltms closed this pull request 2026-09-04 06:14:11 +02:00
Some checks are pending
CI / contract (pull_request) Successful in 1m11s
Details
CI / build (pull_request) Successful in 1m19s
Details

Pull request closed

Please reopen this pull request to perform a merge.
Sign in to join this conversation.
Reviewers
No Reviewers
Labels
Clear labels
blocked
Cannot start until something else lands. The body says what.
 needs-live-proof
Merged and green, but never shown working on the running daemon. Not the same as done.
 ready-to-delegate
Scope, files and acceptance criteria are written. A worker can be briefed from the body alone.
 silent-default
A feature that compiles, passes tests, and ships turned off. Nine recurrences and counting.
No Label
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
agent kevin ltms
No Assignees
2 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: fleet/fleetd#287
Delete Branch "worker/fix-284-backend-error-seat-85912c-11"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?