Wow, Priyanshu, that is a masterclass in reliability. External checkpoints in Postgres + a watchdog cron is bulletproof, but man, that is a massive amount of infrastructure overhead just to make sure n8n isn’t lying to you about a green checkmark!
Really appreciate the detailed breakdown. It confirms exactly why these silent drops are so dangerous if you don’t engineer a whole secondary system to catch them. Thanks for sharing this, super helpful
Ha, fair point — it IS a lot of infrastructure. Honest framing: for hobby workflows, skip it. For production where real money or customer trust is on the line (voice agents calling leads, AI touching financial records, anything with SLAs), the overhead pays for itself the first time it catches a silent drop.
I build it as a reusable pattern — one checkpoint table + one watchdog cron per client, then every workflow for that client gets monitored “for free.” Marginal cost drops fast after the first setup.
Are you working on something specific where this came up, or just comparing notes?
Spot on! And to answer your question — yes, actually, I was asking because my brother and I are working on something specific to solve this exact headache.
We noticed that every serious n8n operator eventually builds the exact same “checkpoint DB + watchdog cron” infrastructure you just described, purely to defeat the green checkmark bug.
So we decided to turn that exact pattern into a lightweight SaaS (we’re calling it NoCrash). The idea is to completely replace that custom infrastructure. You just drop a webhook at the end of a workflow, set an expected schedule, and if NoCrash doesn’t hear from it, it alerts you. No databases to manage, no cron jobs to maintain.
We are just comparing notes with power users right now to make sure we are building it right. Since you’ve literally built this logic from scratch yourself, I’d absolutely love to get your harsh feedback on it once we open the beta (if you’re open to it, of course!).
Nice — that’s actually a really clean productization of the problem. Every senior operator I know has built the same pattern at least three
times. Count me in for the beta when it opens — happy to put it in front of real production workflows and give you harsh honest feedback once
it’s live.
Awesome, I really appreciate it! I will definitely take you up on that offer.
I’ll shoot you a DM here on the forum the second we have the first beta ready for you to tear apart. Thanks again for sharing your workflow insights!