RE: FreeBSD 3.4, Squid 2.3 Stable 2 Reboots server with large files

Actually I had been able to fix this by going from 2.3 S2 to 2.4 Devel 2. Also, I started using the DiskD and it stabled out the Server... FreeBSD 3.4 and UFS/Async-IO just chokes and reboots my server. DiskD on the otherhand, makes it purrr like anything I have ever seen.

I would recommend people who has problems with FreeBSD 3.4 to goto Squid 2.4D2 with Diskd turned on. It saved my insanity :)

-----Original Message-----
From: Clifton Royston [mailto:cliftonr@lava.net]
Sent: Monday, March 13, 2000 2:44 PM
To: Duane Wessels
Cc: Rudy Komsic; squid-users@ircache.net; squid-bugs@ircache.net
Subject: Re: FreeBSD 3.4, Squid 2.3 Stable 2 Reboots server with large
files

On Sat, Mar 11, 2000 at 10:03:10PM -0700, Duane Wessels wrote:
> On Sat, 11 Mar 2000, Rudy Komsic wrote:
>
> > Hello,
> >
> > First off there is a problem with some sort of page fault with Squid
> > cause I got a server crash and dump as the following:
> >
> > Fatal trap 12: page fault while in kernel mode
> >
> > Fault virtual address: 0xc15e5ff0
> > Fault Code = supervisor write, page not present
> > instructor pointer = 0x8:0xc01c2bbc
...
> > trap number = 12
> > panic: page fault
> >
> > This had occurred while downloading a large file. large files can
> > not be downloaded via Squid cause the file usually comes in corrupted
> > and/or never completes properly.
...
> If using squid causes your operating system to crash, then you
> have an operating system bug.
>
> Maybe you should try FreeBSD-3.3?

I've been using a recent version of -3.4-stable with no problems.

Given the nature of the trap (page fault in kernel mode) *and* the
reported associated problems of large files getting corrupted in
transit, I would strongly suspect problems with RAM.

1) If you are overclocking either the processor or RAM (not that you've
said you were!) step it back down to its listed speed and see if that
corrects the problem.

2) Remove each bank of RAM one by one (or swap it with other equivalent
RAM) and see if that corrects the problem.

  -- Clifton

-- 
 Clifton Royston  --  LavaNet Systems Architect --  cliftonr@lava.net
      The named which can be named is not the Eternal named.