OEM12c agents high CPU usage alerts fix (leap second Linux kernel bug)
BrianPaJul 2 2012 — edited Jul 5 2012Just throwing this out here in case anyone else spent the weekend pulling their hair out and hasn't resolved this yet.
Due to the leap second that occurred at midnight UTC going into July 1st, most of my OEM 12c BP1 agents on Linux servers went haywire, taking up 100%+ CPU on monitored servers. This appears to be due to a bug triggered by the leap second.
The following seems to fix the issue without requiring a server reboot.
as root:
/etc/init.d/ntp stop
date `date +"%m%d%H%M%C%y.%S"`
/etc/init.d/ntp start
as OEM agent user:
$AGENT_HOME/bin/emctl stop agent ; $AGENT_HOME/bin/emctl start agent
I didn't discover the fix, but I've confirmed it works.
Edited by: BrianP on Jul 2, 2012 11:53 AM -- subject changed to note this is a Linux kernel bug, not a Java bug... kernel bug causes software using futexes (like Java) to spin and timeout repeatedly