RE: [squid-users] Re. Squid 2.5 with WCCP2 (on Redhat 9)

Henrik,

We decided to go for WCCPv1. I tried to use your module ip_wccp.c which
compiled fine. However, after a few minutes working fine, the server starts
producing this output below. Can you please specify which (in your opinion)
is a working version of Redhat, and which kernel version?

10x

Simon

Oct 24 19:37:41 frodo kernel: recvmsg bug: copied FB63EC97 seq FB63F097
Oct 24 19:37:41 frodo kernel: recvmsg bug: copied 8045ED68 seq 8045F168
Oct 24 19:37:41 frodo kernel: recvmsg bug: copied FB63EC97 seq FB63F097
Oct 24 19:38:07 frodo last message repeated 3975 times
Oct 24 19:40:09 frodo shutdown: shutting down for system reboot
Oct 24 19:40:26 frodo kernel: Unable to handle kernel NULL pointer
dereference at virtual address 00000000
Oct 24 19:40:26 frodo kernel: printing eip:
Oct 24 19:40:26 frodo kernel: c0115473
Oct 24 19:40:26 frodo kernel: *pde = 00000000
Oct 24 19:40:26 frodo kernel: Oops: 0000
Oct 24 19:40:26 frodo kernel: CPU: 0
Oct 24 19:40:26 frodo kernel: EIP: 0010:[<c0115473>] Not tainted
Oct 24 19:40:26 frodo kernel: EFLAGS: 00010016
Oct 24 19:40:26 frodo kernel: eax: 00000000 ebx: f537b800 ecx: 00000001
edx: 00000000
Oct 24 19:40:26 frodo kernel: esi: ecd64364 edi: 00000003 ebp: f591ff7c
esp: f591ff60
Oct 24 19:40:26 frodo kernel: ds: 0018 es: 0018 ss: 0018
Oct 24 19:40:26 frodo kernel: Process shutdown (pid: 747,
stackpage=f591f000)
Oct 24 19:40:26 frodo kernel: Stack: 00000000 c015007e 00000001 00000282
f537b800 f537b864 f537b85c f537b800
Oct 24 19:40:27 frodo kernel: c014ed63 ecd642c0 00000000 f537b800
00000000 00000001 bffffbc8 c014ef06
Oct 24 19:40:27 frodo kernel: f537b800 00000000 c013afa4 00000000
00000000 f591e000 c013afff 00000000
Oct 24 19:40:27 frodo kernel: Call Trace: [<c015007e>] [<c014ed63>]
[<c014ef06>] [<c013afa4>] [<c013afff>]
Oct 24 19:40:27 frodo kernel: [<c01090ff>]
Oct 24 19:40:27 frodo kernel:
Oct 24 19:40:27 frodo kernel: Code: 8b 02 85 c7 75 17 8b 1b 39 f3 75 f1 ff
75 f0 9d 83 c4 10 5b
Oct 24 19:40:58 frodo shutdown: shutting down for system reboot
Oct 24 19:40:58 frodo kernel: <1>Unable to handle kernel NULL pointer
dereference at virtual address 00000000
Oct 24 19:40:58 frodo kernel: printing eip:
Oct 24 19:40:58 frodo kernel: c0115473
Oct 24 19:40:58 frodo kernel: *pde = 00000000
Oct 24 19:40:58 frodo kernel: Oops: 0000
Oct 24 19:40:58 frodo kernel: CPU: 0
Oct 24 19:40:58 frodo kernel: EIP: 0010:[<c0115473>] Not tainted
Oct 24 19:40:58 frodo kernel: EFLAGS: 00010012
Oct 24 19:40:58 frodo kernel: eax: 00000000 ebx: f537b800 ecx: 00000001
edx: 00000000
Oct 24 19:40:58 frodo kernel: esi: e76c62e4 edi: 00000003 ebp: f591ff7c
esp: f591ff60
Oct 24 19:40:58 frodo kernel: ds: 0018 es: 0018 ss: 0018
Oct 24 19:40:58 frodo kernel: Process shutdown (pid: 751,
stackpage=f591f000)
Oct 24 19:40:58 frodo kernel: Stack: 00000000 c015007e 00000001 00000286
f537b800 f537b864 f537b85c f537b800
Oct 24 19:40:58 frodo kernel: c014ed63 e76c6240 00000000 f537b800
00000000 00000001 bffffbc8 c014ef06
Oct 24 19:40:58 frodo kernel: f537b800 00000000 c013afa4 00000000
00000000 f591e000 c013afff 00000000
Oct 24 19:40:58 frodo kernel: Call Trace: [<c015007e>] [<c014ed63>]
[<c014ef06>] [<c013afa4>] [<c013afff>]
Oct 24 19:40:58 frodo kernel: [<c01090ff>]
Oct 24 19:40:58 frodo kernel:
Oct 24 19:40:58 frodo kernel: Code: 8b 02 85 c7 75 17 8b 1b 39 f3 75 f1 ff
75 f0 9d 83 c4 10 5b
Oct 24 19:40:59 frodo shutdown: shutting down for system reboot

-----Original Message-----
From: Henrik Nordstrom [mailto:hno@squid-cache.org]
Sent: Tuesday, October 21, 2003 12:12 PM
To: Simon Vella
Cc: squid-users@squid-cache.org
Subject: Re: [squid-users] Re. Squid 2.5 with WCCP2 (on Redhat 9)

On Tue, 21 Oct 2003, Simon Vella wrote:

> a) The WCCPv2 patch (found at
> http://www.swelltech.com/pengies/joe/patches/squid-2.2.wccp2.patch) should
> work with Squid 2.5, right? Is there another patch for Squid 2.5 to
support
> WCCPv2?

The patch found at http://devel.squid-cache.org/ probably has a higher
likelyhood of working.

> b) The WCCPv2 patch for Squid makes squid talk WCCPv2 with the routers. So
> why do we need the ip_wccp.o module for the kernel?

As WCCPv1, WCCPv2 consists of two protocols

a) WCCPv2 UDP protocol used for announcements between the cache and the
router

b) WCCPv2 GRE protocol used for client traffic forwarded by the router to
the cache. This GRE protocol is different from the normal GRE IP tunneling
protocol and therefore requires a different GRE driver in the kernel. The
ip_wccp module is one such driver. Please note that there exists two
different ip_wccp modules, one for WCCPv1 and one for WCCPv2. It is
important you use the correct one for the WCCP version you are using.

> d) The Linux box must be set up to use ip_tables to redirect the HTTP
> requests received through the GRE tunnel, to the squid port (8080 in our
> case). The destination IP is not changed, so does the Squid still accept
the
> HTTP packet because it is set in HTTP_accelerator mode?

Yes.

Regards
Henrik
Received on Fri Oct 24 2003 - 12:19:35 MDT