Many UNIX agents (linux, solaris, aix, freebsd, openwrt) as well as the mk_oracle agent plugin share the same functionality when running asynchronous/cached plugins:
After they got invoked as subprocesses, they will be watched by subsequent calls of the agent.
If they're still running after twice the cache time, they'll be terminated with .
Previously, the termination has not been verified.
As a result, in situations where a processes couldn't be terminated with , the agent created new processes unlimitedly.
This situation is now mitigated by checking for success after terminating a timed out process.
If the termination failed, no further process of the same plugin will be started.
Additionally, timed out processes, as well as processes that failed to be terminated, will now be reported to the monitoring server.
Affected plugins will be displayed at the "Check_MK Agent" service, alongside a state.
To the list of all Werks