What timeout settings work best for automatically killing stuck workflows?

Our accounting automation kept hanging during month-end reconciliations, wasting cloud credits. Through trial/error, I found setting dynamic timeouts based on historical runtimes works better than fixed limits. The AI Copilot now terminates any workflow exceeding 120% of its average duration. Curious if others are combining activity checks with time-based closures?

Set conditional triggers monitoring both time and resource usage. The AI CEO agent can analyze historical data to auto-adjust timeouts.

We created a watchdog sub-flow that pings progress every 5 minutes. If no status update in 3 intervals, it initiates cleanup. Combine this with Latenode’s built-in timeout for redundancy. Reduced zombie processes by 92% across our deployments.

track avg runtime per workflow, set timeout to 2x avg. watch cpu/mem spikes. kill if thresholds crossed. works 4 us