My system is running solaris zones and oracle dbs in all the zones .. and now i find more then 4500 defunct processes in ps output. Can anyone know about this how to kill these
# ps -ef | grep defunct | wc -l
4899
And when i do ptree on one of the pid it shows most of the defunct processes
originates from one parent thats an oracle db.
#ps -ef | grep defunct | more
adm_itss 11002 3868 0 - ? 0:00 <defunct>
#ptree 11002
3868 ora_psp0_soadb
11002 <defunct>
defunct processes are already dead. you might want to reboot the machine or kill the PARENT process
if your system is not in production, i will suggest you to reboot
Thanx i was able to remove almost all defunct processes
#ps -ef | grep defunct | wc -l
2
But system is still responding slow .. do not know why .. though the mem and cpu stats are following
top
last pid: 8488; load avg: 0.09, 0.11, 0.11; up 58+18:40:39 06:37:36
381 processes: 378 sleeping, 1 zombie, 1 stopped, 1 on cpu
CPU states: 99.7% idle, 0.0% user, 0.3% kernel, 0.0% iowait, 0.0% swap
Memory: 64G phys mem, 41G free mem, 36G total swap, 35G free swap
Can you do a reboot? There are many causes behind the scene as to why servers are slow. If it happens again, all of a sudden, its worth while investigating. Since this is the 1st time it happened, I will be a bit lenient on this. Since there are "things" being done in the system, we can assure whats the ROOT CAUSE for now
And also make sure that the /tmp space is not too full. This kind of symptom has ever occurred, and when the system goes low on swap space, it gets panic. Pls avoid this