OpenClaw
OpenClaw puts a 60-second leash on stalled fallback setup
An open OpenClaw patch expands the cron watchdog from four control transitions to 28 previously uncovered execution-to-setup paths. A fallback model that stalls after a real execution milestone should stop after 60 seconds instead of occupying a cron slot for as long as an hour.
openclaw/openclaw PR #114815 is the inspected primary source: “fix(cron): stop stalled fallback setup across all execution phases.” PR #114815 targets a scheduler path where model fallback can enter setup after several different execution phases. The existing watchdog covered only four before-agent-reply control transitions, leaving 28 execution-to-setup combinations outside the intended 60-second pre-execution limit. The patch replaces the special case with the project’s canonical phase-to-watchdog mapping. Its author reports a 35-of-35 watchdog suite, five repeated timer-regression runs totaling 510 tests, 1,831 cron tests across 155 files, and 166 related Gateway tests. The change claims no schema, configuration, provider, protocol, or stored-job migration. Adjacent PR #114820 separately gives scheduled, startup, and manual runs one durable execution identity so completed jobs do not reappear after a Gateway restart. PR #114829 removes three session-store cloning costs from the catalog path used by the Control UI.
The facts
- openclaw/openclaw PR #114815 was created 2026-07-28T00:22:15Z and updated 2026-07-28T00:57:41Z; state is open. - openclaw/openclaw returned 384351 GitHub stars, 6741 open issues, default branch main, and last push 2026-07-28T00:57:04Z. - Latest tagged-release baseline checked: v2026.7.2-beta.3 published 2026-07-18T23:16:53Z. - PR #114815 targets a scheduler path where model fallback can enter setup after several different execution phases. - The existing watchdog covered only four before-agent-reply control transitions, leaving 28 execution-to-setup combinations outside the intended 60-second pre-execution limit. - The patch replaces the special case with the project’s canonical phase-to-watchdog mapping. - Its author reports a 35-of-35 watchdog suite, five repeated timer-regression runs totaling 510 tests, 1,831 cron tests across 155 files, and 166 related Gateway tests. - The change claims no schema, configuration, provider, protocol, or stored-job migration.
What changed
PR #114815 targets a scheduler path where model fallback can enter setup after several different execution phases. The existing watchdog covered only four before-agent-reply control transitions, leaving 28 execution-to-setup combinations outside the intended 60-second pre-execution limit. The patch replaces the special case with the project’s canonical phase-to-watchdog mapping. Its author reports a 35-of-35 watchdog suite, five repeated timer-regression runs totaling 510 tests, 1,831 cron tests across 155 files, and 166 related Gateway tests. The change claims no schema, configuration, provider, protocol, or stored-job migration. Adjacent PR #114820 separately gives scheduled, startup, and manual runs one durable execution identity so completed jobs do not reappear after a Gateway restart. PR #114829 removes three session-store cloning costs from the catalog path used by the Control UI. The primary record was inspected with its timestamps, state, body and adjacent project records. ClawCharts supplied the assignment list; it does not establish the claim.
Why it matters
Fallback is supposed to improve availability, but an unbounded fallback handshake can consume the very execution slot meant to preserve it. Operators need phase transitions to inherit one timeout contract rather than relying on whichever branch happened to install a timer. The sibling restart fix matters because a scheduler that stops promptly but duplicates completion after restart is still not trustworthy. The useful boundary is where persisted state, visible control and runtime behavior are expected to agree.
Current
All three source records were open at inspection. OpenClaw’s latest tagged baseline was v2026.7.2-beta.3, published July 18; no inspected release establishes that these scheduler changes have shipped. The source therefore supports current mainline movement and test claims, not production behavior in the tagged beta. Repository metadata, current issues and pull requests, release baseline, Hacker News discovery and the rendered ranking row were checked. Weak search residue and historically used source spines were not promoted.
Evidence
The primary URL is https://github.com/openclaw/openclaw/pull/114815. Supporting links are inspected current project records that establish implementation, adjacent direction or boundary conditions. Dates, reproductions, test totals and deployment observations remain attributed to their authors unless this brief explicitly says they were independently rerun.
Source boundary
Open work is described as open, closed work as closed, and operator reports as reports. Closure can reflect a documentation decision or triage outcome; it is not silently translated into a shipped code fix. Ranking explains why the project was scanned; it does not prove correctness, adoption or package behavior.
Operator take
Track reservation, admission, setup start, execution start, terminal state and restart recovery as separate timestamps under one durable run identity. Alert on occupied slots whose phase-specific deadline expired, and test every transition in the phase graph rather than a hand-picked happy path. Preserve a before-state receipt and verify the narrowest invariant where state changes. Do not broaden permissions or delete state merely because the visible symptom is inconvenient.
Caveat
Public project records are mutable. Source state was captured at publication time, and later revisions may change status, scope or evidence. No source here authorizes automatic mutation of a reader’s deployment.
Source inspected; source state, environment and release boundary remain explicit.