Add latent PR worker hibernation lifecycle - #1862
Open
ivannovak wants to merge 3 commits into
Open
Conversation
Owner
|
Automated reminder: thanks for the PR! This branch currently has a merge conflict with the base branch. When you get a chance, please rebase onto (or merge) the latest base branch, resolve the conflict, and push. After that, checks will re-run and the PR will get looked at again. Noted for firstmate#1862 at |
Compose the hibernation lifecycle with main's deferred session-start network stage (kunchenguid#1860). Both sides restructured how session start sequences its work, so the resolution is a composition, not a side pick: - main split bin/fm-bootstrap.sh into a local pass and a network-only pass (FM_BOOTSTRAP_NETWORK_PHASE), moving four network sweeps into the detached bounded worker owned by bin/fm-startup-network.sh and reporting them in the digest's NETWORK CHECKS section. - the hibernation commit added an eighth mutating sweep, latent transaction recovery, written against the pre-deferral ordering. Latent transaction recovery stays on the LOCAL pass, inside main's `local_phase` guard, because fm-latent.sh recover-all reconciles an interrupted hibernation against local refs and the local manifest only and makes no external-network call. Deferring it would buy no startup latency while leaving a half-applied transaction unreconciled beneath the fleet-state digest that reports that task's endpoint. The hibernation ENTER path is the network half of the lifecycle and stays off the session-start path entirely: bin/fm-pr-check.sh drives it when a PR is registered and bin/fm-watch.sh drives it from the supervision poll. Session start therefore still makes no blocking external-network call. Sweep count reconciled from six to seven across AGENTS.md, bin/fm-bootstrap.sh, and bin/fm-session-start.sh so one owner describes the composed ordering, and the bootstrap-diagnostics trigger list is the union of both sides (NETWORK_CHECKS and LATENT_RECOVERY).
The hibernation lifecycle is otherwise network-driven, which makes this one sweep's placement easy to get wrong in exactly two silent ways: wire it to the deferred network half and it is installed but invisible to the fleet-state digest that reports the task's endpoint, or drop it and the whole lifecycle goes inert while still looking present in the tree. recover-all is purely local - it reconciles an interrupted hibernation against local refs and the local manifest - so it belongs on the local half. The test asserts it runs there, that the network half does not repeat it, and that a lock-refused detect-only session does not run it at all. Each assertion was watched failing before this landed: removing the sweep fails the local-half assertion, moving it outside both guards fails the network-repeat assertion, and dropping only the detect-only guard fails the read-only assertion.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
pr_ready_head/pr_headidentity rather than status-log pause tails for eligibilityLocal evidence
bash tests/fm-latent.test.shfailed withnot ok - fm-latent.sh is missingbash tests/fm-latent.test.shpassed every latent lifecycle casebin/fm-lint.shpassed with ShellCheck 0.11.0bin/fm-doc-audience-check.shpassedCI
CI did not run. This is a fork pull request, so workflow runs require maintainer approval. An empty check set is not being reported as passing checks.