ISSUE SUMMARY
MongoDB catches a number of signals and then gracefully shuts down with a log message that mentions the caught signal. The SIGXCPU signal (i.e. running out of CPU time limited by ulimits) is currently not caught and therefore looks identical to a hard kill -9 (SIGKILL) in the log file.
USER IMPACT
It is not clear why the server shut down when it receives a SIGXCPU signal. Diagnosing such an issue is therefore harder than it could be.
SOLUTION
Add the SIGXCPU signal to the list of caught signals and quit gracefully with a log message.
WORKAROUNDS
None
AFFECTED VERSIONS
All recent production releases up to version 2.4.9 are affected.
PATCHES
The fix is included in the 2.4.10 production release and the 2.5.5 development version, which will evolve into the 2.6.0 production release.
Original Description
From a logfile, it's impossible to tell the difference between a hard kill -9, and running out of cpu time (ie. ulimits), which generates a sigxcpu. Adding this signal to the list of caught signals would help with diagnosing this problem.