On 23/12/2011 4:53 p.m., Nguyen Hai Nam wrote:
Hi,
Today I've got this message from Squid:
"You've encountered a fatal error in the Squid Cache version 3.2.0.14.
If a core file was created (possibly in the swap directory),
please execute 'gdb squid core' or 'dbx squid core', then type 'where',
and report the trace back to squid-bugs@xxxxxxxxxxxxxxx."
This message appeared after squid process bump to 100% CPU load and I
have to kill its PID to restart.
The traffic is normal news website and/or youtube, etc ...
I'm running intercept squid (with IPfilter NAT) on OpenIndiana 151a
system, how can I trace this error?
The message is created when the kill signal is received. In case it was
an automatic kill. The core and message are usually not much use when
done manually.
Although in this case the problem actually could be related to what
Squid was doing at the time of kill. If a core exists you can use a
debugger (ie gdb) to load the core and see what Squid was doing when
killed. Otherwise you will need to wait for teh CPU to reach 100% again
and run strace on the squid worker processes PID to find out what is
sucking up the CPU cycles.
Amos
