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

CB-544: shutdown drain preserves worker worktrees — no data loss on daemon restart #24

Closed
agent wants to merge 0 commits from worker/cb-544-drain-preserves-worktree-925fad-3 into main
pull from: worker/cb-544-drain-preserves-worktree-925fad-3
merge into: fleet:main
Conversation 1 Commits - Files Changed -
agent commented 2026-08-13 20:37:51 +02:00
Member
Copy Link
Copy Source

Fixes confirmed data-loss defect CB-544: SessionManager.drainAll() (shutdown drain) fused two concerns in release() — stopping the worker pane and removing its worktree. A daemon restart while a worker was mid-turn deleted that worker's worktree (the sole copy of any uncommitted work) after a 5s default drain timeout.

What changed:

  • release() now takes an explicit ReleaseCause (COMPLETED vs SHUTDOWN). A COMPLETED release (explicit bridge_stop / REST stop / reaper / context-cap / recycle) still stops the pane AND removes the worktree — unchanged. A SHUTDOWN release (drainAll on close) stops the pane but PRESERVES the worktree.
  • drainAll() releases with SHUTDOWN. A session still BUSY when the drain timeout expires is abandoned mid-turn and logged loudly (WARN) with the preserved worktree path so an operator can find and reclaim it. Idle sessions get an INFO line noting the preserved path.
  • Orphaned worktrees are acceptable and strictly preferable to destroyed work; no automatic cleanup is added (reclamation left to boot reconciliation).

Tests (added to WorktreeSessionManagerTest against the existing FakeWorktrees seam):

  • drainAll preserves the worktree of an idle session while still stopping its pane
  • drainAll preserves the worktree of a session still BUSY at timeout while still stopping its pane
  • Existing completed-teardown (releaseRemovesWorktreeButDoesNotDeleteBranch), reaper, and drain tests pass unchanged.

Verified: mvn clean install — Tests run: 542, Failures: 0, Errors: 0, Skipped: 0 — BUILD SUCCESS (based on origin/main 91c9f98).

Fixes confirmed data-loss defect CB-544: SessionManager.drainAll() (shutdown drain) fused two concerns in release() — stopping the worker pane and removing its worktree. A daemon restart while a worker was mid-turn deleted that worker's worktree (the sole copy of any uncommitted work) after a 5s default drain timeout. What changed: - release() now takes an explicit ReleaseCause (COMPLETED vs SHUTDOWN). A COMPLETED release (explicit bridge_stop / REST stop / reaper / context-cap / recycle) still stops the pane AND removes the worktree — unchanged. A SHUTDOWN release (drainAll on close) stops the pane but PRESERVES the worktree. - drainAll() releases with SHUTDOWN. A session still BUSY when the drain timeout expires is abandoned mid-turn and logged loudly (WARN) with the preserved worktree path so an operator can find and reclaim it. Idle sessions get an INFO line noting the preserved path. - Orphaned worktrees are acceptable and strictly preferable to destroyed work; no automatic cleanup is added (reclamation left to boot reconciliation). Tests (added to WorktreeSessionManagerTest against the existing FakeWorktrees seam): - drainAll preserves the worktree of an idle session while still stopping its pane - drainAll preserves the worktree of a session still BUSY at timeout while still stopping its pane - Existing completed-teardown (releaseRemovesWorktreeButDoesNotDeleteBranch), reaper, and drain tests pass unchanged. Verified: mvn clean install — Tests run: 542, Failures: 0, Errors: 0, Skipped: 0 — BUILD SUCCESS (based on origin/main 91c9f98).
agent added 1 commit 2026-08-13 20:37:52 +02:00
CB-544: shutdown drain preserves worker worktrees (no data loss)
CI / contract (pull_request) Successful in 41s
Details
CI / build (pull_request) Successful in 1m29s
Details
04e21c9243
ltms commented 2026-08-15 07:10:54 +02:00
Owner
Copy Link
Copy Source

Closing as landed. Verified with git cherry main <branch>: every commit's content is present in main. The branch tip is not an ancestor only because the commit was rebased before merge, so the SHA differs while the patch is identical. No code is lost. Verified 2026-08-15.

Closing as **landed**. Verified with `git cherry main <branch>`: every commit's content is present in `main`. The branch tip is not an ancestor only because the commit was rebased before merge, so the SHA differs while the patch is identical. No code is lost. Verified 2026-08-15.
ltms closed this pull request 2026-08-15 07:10:56 +02:00
Some checks are pending
CI / contract (pull_request) Successful in 41s
Details
CI / build (pull_request) Successful in 1m29s
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#24
Delete Branch "worker/cb-544-drain-preserves-worktree-925fad-3"

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?