Werk #22466: Relay: Core accepts results from many relay hosts at once
| Component | Relay | ||
| Title | Relay: Core accepts results from many relay hosts at once | ||
| Date | Oct 29, 2026 | ||
| Level | Trivial Change | ||
| Class | Bug Fix | ||
| Compatibility | Compatible - no manual interaction needed | ||
| Checkmk versions & editions |
|
Previously, relays with many hosts lost part of their monitoring results. For example, one relay with 404 SNMP hosts got 148 of its 404 results per minute through. The rest were dropped, and services of relay hosts went stale. Adding more relays didn't help, because all results take the same path into the core.
There were two causes. The monitoring core accepted relay results over a
single connection at a time. When that connection was busy, the agent
receiver gave up at once and answered the relay with HTTP 502 instead of
waiting. You could see this as POST /relays/<id>/monitoring requests
with status 502 in var/log/agent-receiver/access.log.
Both are fixed. The core now accepts results over many connections in parallel. When the core is busy, the agent receiver now waits up to 5 seconds for it.