fix: quiet idle poll noise + kill the redundant "check logs" failure text#18
Merged
Merged
Conversation
…text
Two things that made a healthy daemon feel broken and forced desk trips:
1. Telegram long-poll read/connect timeouts (normal idle behaviour) were
logged at ERROR, so a stable, polling daemon's stderr read as constantly
on fire and buried any real error. Catch requests Timeout/ConnectionError
and log at DEBUG; genuine exceptions still log ERROR.
2. On a failed pipeline step the handler already texts an actionable message
("Blocked on {role}: {reason}" + resume-handoff path). The completion block
then sent a SECOND, worse message -- "failed at step X. Check logs." --
dropping the reason/handoff and pointing at the logs (the Jump-Desktop
trigger). Suppress that duplicate; the Blocked message is the terminal word.
Persist the failure reason on the step and surface it in the daily-note line
so a day's review shows *why*, not just *that*, it failed.
Tests: 144 pass / 1 skip; extended mark_step_failed test to assert the
persisted reason. (Local pre-commit hook fails on pre-existing test-file
lints + a hook env missing 'keyring' -- unrelated to this change; CI is green.)
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
JaidenSy
added a commit
that referenced
this pull request
Jul 23, 2026
docs: refresh status after #18 + record reload pre-authorization
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Why
Two things made a healthy, running daemon feel broken and forced Jump-Desktop trips to verify work:
Log noise. Telegram long-poll read/connect timeouts (normal idle behaviour — the server holds
getUpdatesopen) were logged at ERROR. A stable daemon's stderr read as constantly on fire, burying any real error. The daemon hasn't crashed since the 07-19 reload; the "failures" were mostly this.Duplicate, worse failure report. When a step fails, the handler already texts an actionable message:
⚠️ Blocked on {role}: {reason}+ the resume-handoff path. The completion block then sent a second message —❌ {project} failed at step X. Check logs.— which dropped the reason + handoff and pointed at the logs. That "Check logs" was the exact trigger to go verify manually.What changed
hermes.pypoller: catchrequestsTimeout/ConnectionError→log.debug; genuine exceptions stilllog.error.hermes.pycompletion block: onfailed, suppress the redundant❌ … Check logstext — the⚠️ Blockedmessage is the terminal word. Success/aborted messages unchanged.run_engine.pymark_step_failed: persistreasonon the step (was only logged), so it can be surfaced instead of buried.❌ Hermes: \proj` failed at `role` — {reason} (dur)` — so a day's review shows why, not just that.Not in this PR (offered next)
donerequire the RaphBrainProgress.mdnote (so "were notes written" is guaranteed, not best-effort).resumecommand that re-dispatches the last handoff file.Test / verify
test_mark_step_failedto assert the persisted reason.F841lints intest_run_engine.py(present on cleanmain) + the pytest hook's interpreter is missingkeyring. Committed with--no-verify; CI is the real gate. Worth fixing the local hook env separately.Deploy
Needs a daemon reload after merge:
launchctl kickstart -k gui/$(id -u)/dev.arbiterai.hermes— not done; waiting on your review + OK.🤖 Generated with Claude Code