Summary

Cron RPC queries (cron.list, cron.status) timeout when a job with wakeMode: "now" and sessionTarget: "main" is executing, because executeJob() holds the lock while awaiting runHeartbeatOnce().

Reproduction

1. Create a cron job with wakeMode: "now" and sessionTarget: "main"
2. Wait for the job to fire
3. While the agent is processing the triggered turn, call cron list
4. Query times out (10s default)

Root Cause

In dist/cron/service/timer.js lines 67-87:

// Inside executeJob(), which runs inside locked()
if (job.wakeMode === "now" && state.deps.runHeartbeatOnce) {
    for (;;) {
        heartbeatResult = await state.deps.runHeartbeatOnce({ reason });
        if (...) break;
        await delay(250);  // Loops up to 2 minutes
    }
}

The locked() function in ops.js serializes ALL cron operations through state.op. Since executeJob() awaits the agent turn completion while holding the lock, any concurrent RPC query blocks.

Evidence from Logs

⇄ res ✓ cron.list 34021ms  // Query took 34s (waited for lock)
⇄ res ✓ cron.status 15254ms // 15s wait

Suggested Fix

Options:

1. Release lock before awaiting heartbeat - update job state after heartbeat completes
2. Separate read/write locks - allow concurrent reads during job execution
3. Move heartbeat await outside locked() - restructure to not hold lock during waits

Environment

• Clawdbot: 2026.1.24-3
• Node: v25.1.0
• macOS (arm64)