squid randomly dying (FreeBSD 2.2.5)

From: Jordyn A. Buchanan <jordyn@dont-contact.us>
Date: Tue, 9 Dec 1997 00:12:30 -0500

Hello:

I've been running a squid cache for many months now, generally without any
sort of problems. Recently, however, I upgraded the system that it is
running on to FreeBSD 2.2.5 (from 2.1.7) and the problems began...

Now, the squid process suddenly dies every 18-36 hours, sometimes slightly
more frequently. In my log files, I generally see something like this
right before it dies:

>97/12/08 23:58:06| reconfigure: SIGHUP received
>97/12/08 23:58:06| Waiting 30 seconds for active connections to finish
>97/12/08 23:58:06| FD 21 Closing HTTP connection
>97/12/08 23:58:06| FD 22 Closing ICP connection

I've already taken the following steps to try to resolve the problem
without avail:

1) Upgraded from squid 1.1.9 to 1.1.17 to 1.1.18.
2) Disabling TCP extensions per the FAQ.
3) unlimiting the spawning shell before running squid--occasionally when
   run from the bootup script we would get xmalloc errors, but they seem
   to be unrelated to the problem described above. Running the proccess
   from an unlimited shell seems to help a bit, but only adds a few hours
   to the time before squid craps out. This may be coincidence.
4) I've scoured the system for any process that may be sending squid a kill -1.
   As far as I can tell, there is nothing.
   a) When I actually use kill -HUP on the squid process, I see a very
      similar message to the one above, but squid does not die, but actually
      reconfigures like it is supposed to.
   b) There is no evidence in the logs of HUP signals being received at any
      time other than when I am intentionally sending it and when it is dying.
5) Installed all relevant FreeBSD security patches. At one point I thought
   this might be part of some active attack, but the times are exceedingly
   erratic and I'm not aware of any exploits that could cause this.

If anyone has any suggestions about this, I'd greatly appreaciated.
Babysitting this thing has become a pain in the butt, and even though it's
fairly simple to write a script to simply restart the process when it dies
off (isn't RunCache supposed to do this?), I'd rather figure out what was
going on.

Your help is appreciated,

Jordyn

|----------------------------------------------------------------|
|Jordyn A. Buchanan mailto:jordyn@bestweb.net |
|Bestweb Corporation http://www.bestweb.net |
|Senior System Administrator +1.914.271.4500 |
|----------------------------------------------------------------|
Received on Mon Dec 08 1997 - 21:21:15 MST

This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:37:52 MST