Photo: free stock photography (Unsplash licence) — see imprint
Ping is not monitoring
A server can happily return 200 on the homepage while the payment provider integration throws on every checkout. Your monitoring must answer the only question that matters: are orders still being placed?
Alert on the business metric
Orders per hour, compared to the same hour last week. If it drops by 80% at 10 a.m. on a Tuesday, something is broken — even if every server is green. This single alert has caught more real outages for us than every CPU graph combined.
Errors need an owner and a threshold
An error tracker that nobody reads is a very expensive log file. Route each error type to a person, and set a threshold that means something — 'more than five checkout exceptions in ten minutes' beats 'any error, ever'.
Watch the integrations hardest
Your shop is rarely the thing that fails. It is the ERP that stopped accepting orders, the payment API that changed a field, the certificate that expired on a Sunday. Monitor every outbound integration with its own heartbeat.
- Alert on orders per hour, not just CPU.
- Every alert needs an owner.
- Integrations fail more often than your shop does.
We do this for a living — Shopware, Node.js, React, ERP integration and automation for B2B.
Talk to an engineer