OS version: suse linux enterprise server 12 sp2apache ignite(2.8.0) service(one process with a child process,and the child serve out caculation ability and consumes memorys hard) running on this machine, always be killed accidently.sometimes in two weeks after last startup , sometimes only 3 hours after last startup.(there's other processes on this machine, eg. zookeeper.)
i've got some useful info , but not enough:
NOT oomkiller. no exception in ignite service log and /var/log/message. ignite dissapeared without any exception log, seem that it was killed with signal 9.
i insmod a kernel module to record who is the killer, and found that is PID 1 killed the child process with signal 0 ( the father process soon disappeared , without any killer info). Also, i printed ps -ef every 10 seconds to logs, found that the child process name became "defunct" before it disappeared.
mem and swap were not in starvation when ignite was killed, very smooth and steady before killed. I am very sure with monitoring chart and DIY logs.
I don't know how to go on with it.. helppppp...
