Problem
There are log Log lines , that are missing from the event console but present in the monitored log file.
The configured patterns and filters should allow them to be forwarded to the event console.
Additionally, this behavior is not easily reproducible and happens at random times.
Reason
The reason for this is a certain particular constellation of configuration options.
Per default, the periodic service discovery uses cached agent output stored on the Checkmk server for 120 seconds, ; after that amount of time , it will fetch new data from the host.
If the default check interval for the agent is equal to or smaller than that 120 seconds (which it is by default: 60 seconds), then this error can not occur.
If the normal check interval for the Checkmk agent is larger more significant than 120 seconds, it can come to the following situation:
- The agent is queried normally, and the agent output is cached on the Checkmk server
- The periodic service discovery runs at their configured interval and 120 seconds or more after the agent was queried
- The periodic service discovery considers the cached data too old and fetches a fresh agent output
- This output contains new log lines but is never processed by the logwatch check plugin and therefore is not forwarded to the event console.
- When the agent is queried on the normal regular interval again, it can not realize , that log messages have been fetched but not been processed.
Solution
There are two main solutions:
- Decrease the normal check interval for the Checkmk agent below 120 seconds
Manually change the affected site's sites
main.mk
and add the following lines. Adapt the$SECONDS
to be at least the normal check interval of the Checkmk agent!Warning This is not supported by CheckmkCheckmk does not support this! Do this at your own risk!
Code Block language bash theme RDark max_cachefile_age = $SECONDS cluster_max_cachefile_age = $SECONDS
...