So here were the final instructions for getting a core file. 1. added ulimit -c unlimited to the init.d script. 2. installed debug squid rpm package. 3. echo 1 > /proc/sys/kernel/core_uses_pid 4. echo /corefiles/core-%e-%p-%t > /proc/sys/kernel/core_pattern Now I have the file: -rw------- 1 squid squid 5.5G Dec 24 02:33 core-squid-21428-1261639928 All that's left is to run the through gdb. I'll have more info. soon on what my crash is looking like. Thanks Dusten On Wed, Dec 23, 2009 at 11:21, Dusten Splan <dsplan@xxxxxxxxxxxxxx> wrote: > I just noticed that my ulimit for core dumps was set to 0. I have > changed it to unlimited and I hope to see a core file on the next > crash. > > Just as a note so if others have the same issue. > > To see the ulimit settings > ulimit -a > > To set the core limit > ulimit -c unlimited > > Dusten > > On Wed, Dec 23, 2009 at 03:41, Kinkie <gkinkie@xxxxxxxxx> wrote: >> On Tue, Dec 22, 2009 at 8:54 PM, Dusten Splan <dsplan@xxxxxxxxxxxxxx> wrote: >>> I'm seeing my squid processes die and restart. I think it's related >>> to a bad cache file but I am unsure how to track it down. I have >>> rebuilt the cache by removing the swap.state file and starting squid >>> back up. >>> >>> Here's what I'm seeing in my logs. >>> >>> 2009/12/21 22:07:01| WARNING: 1 swapin MD5 mismatches >>> 2009/12/21 22:07:01| could not parse headers from on disk structure! >>> FATAL: Received Segment Violation...dying. >>> 2009/12/21 22:07:01| storeDirWriteCleanLogs: Starting... >>> 2009/12/21 22:07:01| WARNING: Closing open FD 58 >>> 2009/12/21 22:07:01| WARNING: Closing open FD 59 >>> 2009/12/21 22:07:01| 65536 entries written so far. >>> 2009/12/21 22:07:01| 131072 entries written so far. >> >> Please see if you can find a core dump in your coredump_dir. It might >> contain valuable information to understand the cause of the problem. >> >> Please see http://wiki.squid-cache.org/SquidFaq/BugReporting >> >> >> -- >> /kinkie >> >
