Re: [squid-users] SQuid 2.6Stable10 diskd problems

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On Sun, 11 Mar 2007 15:13:52 +0100
"Robert" <vipit@wp.pl> wrote:

> hello
> i upgraded squid to version 2.6S10 and seen in logs that there is a
> problem with this relase in diskd, squid crashed twice with error:
> 2007/03/06 15:17:48| assertion failed: client_side.c:4209: "buf !=
> NULL || !conn->body.request"
>
> logs looks like:
>
> 2007/03/06 15:17:27| WARNING: suspicious CR characters in HTTP header
> {Pragma: no-cache^MContent-Length: 313}
> 2007/03/06 15:17:27| ctx: enter level 0:
> 'http://status.gadu-gadu.pl/users/status.asp?id=10003644&styl=1'
> 2007/03/06 15:17:27| WARNING: suspicious CR characters in HTTP header
> {Pragma: no-cache^MContent-Length: 313}
> 2007/03/06 15:17:27| ctx: exit level 0
> 2007/03/06 15:17:27| WARNING: suspicious CR characters in HTTP header
> {Pragma: no-cache^MContent-Length: 313}
> 2007/03/06 15:17:27| ctx: enter level 0:
> 'http://status.gadu-gadu.pl/users/status.asp?id=4172805&styl=1'
> 2007/03/06 15:17:27| WARNING: suspicious CR characters in HTTP header
> {Pragma: no-cache^MContent-Length: 313}
> 2007/03/06 15:17:27| ctx: exit level 0
> 2007/03/06 15:17:27| WARNING: suspicious CR characters in HTTP header
> {Pragma: no-cache^MContent-Length: 313}
> 2007/03/06 15:17:27| ctx: enter level 0:
> 'http://status.gadu-gadu.pl/users/status.asp?id=4577990&styl=1'
> 2007/03/06 15:17:27| WARNING: suspicious CR characters in HTTP header
> {Pragma: no-cache^MContent-Length: 313}
> 2007/03/06 15:17:27| ctx: exit level 0
> 2007/03/06 15:17:27| WARNING: suspicious CR characters in HTTP header
> {Pragma: no-cache^MContent-Length: 313}
> 2007/03/06 15:17:46| httpReadReply: Request not yet fully sent "POST
> http://x3.uploaded.to/Download.php?id=246c30cc98c5e28ea0b53b0b2fcc5120"
> 2007/03/06 15:17:48| assertion failed: client_side.c:4209: "buf !=
> NULL || !conn->body.request"
> 2007/03/06 15:18:01| Starting Squid Cache version 2.6.STABLE10 for
> i686-pc-linux-gnu...
> 2007/03/06 15:18:01| Process ID 11962
> 2007/03/06 15:18:01| With 8192 file descriptors available
> 2007/03/06 15:18:01| Using epoll for the IO loop
> 2007/03/06 15:18:01| DNS Socket created at 0.0.0.0, port 33675, FD 5
> 2007/03/06 15:18:01| Adding nameserver 127.0.0.1 from /etc/resolv.conf
> 2007/03/06 15:18:01| Adding nameserver 193.110.121.20
> from /etc/resolv.conf 2007/03/06 15:18:01| Unlinkd pipe opened on FD
> 10 2007/03/06 15:18:01| Swap maxSize 6144000 KB, estimated 472615
> objects 2007/03/06 15:18:01| Target number of buckets: 23630
> 2007/03/06 15:18:01| Using 32768 Store buckets
> 2007/03/06 15:18:01| Max Mem size: 262144 KB
> 2007/03/06 15:18:01| Max Swap size: 6144000 KB
> 2007/03/06 15:18:01| Local cache digest enabled; rebuild/rewrite
> every 3600/3600 sec
> 2007/03/06 15:18:01| Store logging disabled
> 2007/03/06 15:18:01| Rebuilding storage in /cache/cache1 (DIRTY)
> 2007/03/06 15:18:01| Rebuilding storage in /cache/cache2 (DIRTY)
> 2007/03/06 15:18:01| Rebuilding storage in /cache/cache3 (DIRTY)
> 2007/03/06 15:18:01| Using Round Robin store dir selection
> 2007/03/06 15:18:01| Set Current Directory to /cache/cache1
> 2007/03/06 15:18:01| Loaded Icons.
> 2007/03/06 15:18:01| Accepting transparently proxied HTTP connections
> at 0.0.0.0, port 3128, FD 18.
> 2007/03/06 15:18:01| Ready to serve requests.
> 2007/03/06 15:18:04| httpReadReply: Excess data from "HEAD
> http://frog10b.gaduradio.pl/rock"
> 2007/03/06 15:18:14| Store rebuilding is 4.2% complete
> 2007/03/06 15:18:18| WARNING: newer swaplog entry for dirno 0, fileno
> 0000003A
> 2007/03/06 15:18:18| WARNING: newer swaplog entry for dirno 0, fileno
> 000000A8
> 2007/03/06 15:18:18| WARNING: newer swaplog entry for dirno 2, fileno
> 000000A2
> 2007/03/06 15:18:18| WARNING: newer swaplog entry for dirno 2, fileno
> 00000680
> 2007/03/06 15:18:18| WARNING: newer swaplog entry for dirno 0, fileno
> 00000602
> 2007/03/06 15:18:18| WARNING: newer swaplog entry for dirno 0, fileno
> 00000603
> 2007/03/06 15:18:18| WARNING: newer swaplog entry for dirno 0, fileno
> 0000074F
>
> and then cache is corrupted and is visible site browsing slow down,
> cpu uasge i more than normal, i must remove all cache dirs and create
> it again with squid -z.
> this is my problem or squid relase problem ?
>
> Greetz
> Robert
>
>

Hi Robert,

What I think is that this may not be an issue with diskd only. It might
be an issue with Squid-2.6.Stable10.

I am also facing random crashes in Squid-2.6.Stable10 in my
FreeBSD-6.1 server. Luckily Squid restarts itself immediately after
crashing and nobody notices any negative performance.

But still the cache never gets to stabilize itself. I am using COSS and
this exact setup works perfectly in other FreeBSD-6.0 server running
Squid-2.6.Stable9.

Also, since Squid crashes randomly, I am afraid that it might corrupt
it's cache resulting in poor performance.

I also get the following error when Squid crashes:

2007/03/12 10:06:37| assertion failed: client_side.c:4164: "buf != NULL
|| !conn->body.request" 2007/03/12 10:06:41| Starting Squid Cache
|| version 2.6.STABLE10 for i386-unknown-freebsd6.1...
2007/03/12 10:06:41| Process ID 32331
2007/03/12 10:06:41| With 14720 file descriptors available
2007/03/12 10:06:41| Using kqueue for the IO loop
2007/03/12 10:06:41| DNS Socket created at 0.0.0.0, port 59219, FD 5

- --

With best regards and good wishes,

Yours sincerely,

Tek Bahadur Limbu

(TAG/TDG Group)
Jwl Systems Department

Worldlink Communications Pvt. Ltd.

Jawalakhel, Nepal

http://www.wlink.com.np
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.2.2 (FreeBSD)

iD8DBQFF9PDhVrOl+eVhOvYRAu5BAKCENtxe15G5dKgVaD7lRe1uIokD6ACfV2+F
sJpoWZFPFNntvvb0x6hby9U=
=6HU4
-----END PGP SIGNATURE-----
Received on Mon Mar 12 2007 - 00:20:05 MDT