Re: [squid-users] squid 2.6-rproxy: assertion store_client.c:211: "sc->cmp_offset == copy_offset"

The rproxy patch is not currently full up to date to match the
storeClientCopy changes in HEAD.. You are fully correct that the
assertion is parameter checking. The rproxy branch is slowly being
updated when there is time to match the changes, and I was not aware
it could compile in the state it is now..

To tell exacly where the problem is you need to get a stacktrace using
GDB.

If you know some C and wish to help with updating rproxy to match HEAD
please contact me in private.

Regards
Henrik Nordström
MARA Systes AB, Sweden
Author of the rproxy branch

On Sunday 09 June 2002 23:27, Jeffrey D. Wheelhouse wrote:
> I get this error every couple of hours in squid_cache:
>
> 2002/06/09 10:31:42| assertion failed: store_client.c:211:
> "sc->cmp_offset == copy_offset"
> 2002/06/09 12:02:05| assertion failed: store_client.c:211:
> "sc->cmp_offset == copy_offset"
> 2002/06/09 16:24:01| assertion failed: store_client.c:211:
> "sc->cmp_offset == copy_offset"
> 2002/06/09 16:31:12| assertion failed: store_client.c:211:
> "sc->cmp_offset == copy_offset"
> 2002/06/09 18:45:39| assertion failed: store_client.c:211:
> "sc->cmp_offset == copy_offset"
> 2002/06/09 19:34:49| assertion failed: store_client.c:211:
> "sc->cmp_offset == copy_offset"
>
> squid restarts itself and goes about its business, but I would
> prefer to eliminate this.
>
> This occurs with squid built from the current rproxy tag on OpenBSD
> 3.0 and FreeBSD 4.6-RC.
>
> I checked the source and this assertion looks like parameter
> checking in storeClientCopy(). I don't really understand what the
> cmp_offset property is used for or what this test is checking.
> Perhaps this is related to freshness checking (cmp_) or range
> handling (_offset) in some way?
>
> I've correlated squid_access, and there doesn't seem to be any
> particular pattern. The last request before the crash is generally
> a small graphic, but I've had both hits and misses leading up to
> it. It does not appear load related. Today's peak load on this
> machine was after the crash at 12:02 and had abated by 16:24.
>
> Web searches and a scan of the mailing list archives didn't turn up
> anything on this.
>
> Is this a known problem with the rproxy/2.6 branch? If not, what
> can I do to help debug it? I'm attempting to collect a core file
> now. I'm also happy to run it under gdb if I know what to look for
> once I catch the assert.
>
> Thanks very much for any pointers!
>
> Jeff
Received on Mon Jun 10 2002 - 02:11:58 MDT