Hello,
I'm begining a new thread.
To remember, I'm trying to migrate a squid server from 3.1.19 to 3.2.1. The old
one works fine, but with the new version, squid dies each 1 to 3 minutes, with
"segment violation".
I'm not able do get core files under Solaris, despite of correct settings
(ulimit, ...).
So, I set up debug_options to "ALL,8" and "ALL,9". and I wrote a single script
which prints the last 25 lines before each SEGFAULT received.
The result is at http://www.j-chkmail.org/users/squid
The contents of the three files are :
* cores-4.txt - debug_options ALL,8 - squid doing cache
* cores-5.txt - debug_options ALL,9 - squid doing cache
* cores-9.txt - debug_options ALL,8 - cache_dir option commented
(If you want, I can send you the files.)
The third test was to be sure that the problem didn't come from the cache
partition on a zfs filesystem.
Before each "segfault" there is a line indicating the destination hostname. Most
of the time, but not allways, the hostname doesn't resolve or it's a CNAME.
Thanks for your help,
José-Marcio
--
