Mailing List Archive

So sorry for bothering you.

I can see lots of messages like below at /var/log/messages.
the error writes probably every one minute..

Aug 30 17:56:50 www kernel: varnishd[6039]: segfault at 0000000000000000
rip 0000003111a6ff70 rsp 000000004d817c48 error 4
Aug 30 17:57:48 www kernel: varnishd[6068]: segfault at 0000000000000000
rip 0000003111a6ff70 rsp 0000000043002c48 error 4
Aug 30 17:58:49 www kernel: varnishd[6150]: segfault at 0000000000000000
rip 0000003111a6ff70 rsp 0000000055827c48 error 4
Aug 30 17:59:49 www kernel: varnishd[6163]: segfault at 0000000000000000
rip 0000003111a6ff70 rsp 0000000046008c48 error 4
Aug 30 18:00:52 www kernel: varnishd[6210]: segfault at 0000000000000000
rip 0000003111a6ff70 rsp 0000000044004c48 error 4
Aug 30 18:01:50 www kernel: varnishd[6303]: segfault at 0000000000000000
rip 0000003111a6ff70 rsp 0000000053823c48 error 4
Aug 30 18:02:51 www kernel: varnishd[6316]: segfault at 0000000000000000
rip 0000003111a6ff70 rsp 0000000044004c48 error 4
Aug 30 18:03:51 h-cdn04 kernel: varnishd[6381]: segfault at
0000000000000000 rip 0000003111a6ff70 rsp 0000000046008c48 error 4


>From: "Monty Ree" <chulmin2 at hotmail.com>
>To: varnish-misc at projects.linpro.no
>Subject: why purge makes child process restart?
>Date: Thu, 30 Aug 2007 09:02:03 +0000
>
>Hello, list.
>
>my varnishd is 1.1.1 and I found that some problem at varnish.
>When I purge using squidclient like below which in squid package,
>varnishd child process restarted and all caches are deleted.
>
>squidclient -m purge -r -p 80 -h example.com http://example.com/test.jpg
>
>Please confirm about this..
>
>additionally, my system is linux kernel 2.6.18(x86_64) and varnishd1.1.1,
>and child process restarted so requently(about once per 5 minutes) without
>purge.
>
>
>Thanks for your help.
>
>_________________________________________________________________
>?? ???? ??? MSN SMS ???? ?????.
>http://im.msn.co.kr/new/function/function_01_02.asp
>
>_______________________________________________
>varnish-misc mailing list
>varnish-misc at projects.linpro.no
>http://projects.linpro.no/mailman/listinfo/varnish-misc

_________________________________________________________________
?? ??? ?? ???? ?? ???!
http://match.kr.msn.com/channel/index.aspx?trackingid=1002127

"Monty Ree" <chulmin2 at hotmail.com> writes:
> Aug 30 17:56:50 www kernel: varnishd[6039]: segfault at 0000000000000000
> rip 0000003111a6ff70 rsp 000000004d817c48 error 4

A stack trace would be nice...

DES
--
Dag-Erling Sm?rgrav
Senior Software Developer
Linpro AS - www.linpro.no

Hello, all.

I found that varnishd child process auto restarted,
when "Host:" field is not exist at request header.

Please confirm about this..


Thanks for your help..


>From: Dag-Erling Sm?grav <des at linpro.no>
>To: "Monty Ree" <chulmin2 at hotmail.com>
>CC: varnish-misc at projects.linpro.no
>Subject: Re: why purge makes child process restart?
>Date: Thu, 30 Aug 2007 11:16:01 +0200
>
>"Monty Ree" <chulmin2 at hotmail.com> writes:
> > Aug 30 17:56:50 www kernel: varnishd[6039]: segfault at
0000000000000000
> > rip 0000003111a6ff70 rsp 000000004d817c48 error 4
>
>A stack trace would be nice...
>
>DES
>--
>Dag-Erling Sm?grav
>Senior Software Developer
>Linpro AS - www.linpro.no

_________________________________________________________________
MSN Messenger? ?? ????? ?? ??? ??? ????.
http://www.msn.co.kr/messenger

You're right, I'll fix.

--
Poul-Henning Kamp | UNIX since Zilog Zeus 3.20
phk at FreeBSD.ORG | TCP/IP since RFC 956
FreeBSD committer | BSD since 4.3-tahoe
Never attribute to malice what can adequately be explained by incompetence.

Thanks for your help..

Can you give us temporary vcl or urgent patch which drop the GET, HEAD or
purge request which is not including Host header?


Thanks again..


>From: "Poul-Henning Kamp" <phk at phk.freebsd.dk>
>To: "Monty Ree" <chulmin2 at hotmail.com>
>CC: des at linpro.no, varnish-misc at projects.linpro.no
>Subject: Re: why purge makes child process restart?
>Date: Fri, 31 Aug 2007 07:00:07 +0000
>
>You're right, I'll fix.
>
>--
>Poul-Henning Kamp | UNIX since Zilog Zeus 3.20
>phk at FreeBSD.ORG | TCP/IP since RFC 956
>FreeBSD committer | BSD since 4.3-tahoe
>Never attribute to malice what can adequately be explained by incompetence.

_________________________________________________________________
?? ?? ?? ??? ?????? MSN ???? ?????.
http://fortune.msn.co.kr/

In message <BLU116-F54FAED96DBC408BD5A51E85CF0 at phx.gbl>, "Monty Ree" writes:
>Thanks for your help..
>
>Can you give us temporary vcl or urgent patch which drop the GET, HEAD or
>purge request which is not including Host header?

This is what you need:

http://varnish.projects.linpro.no/changeset/1930

--
Poul-Henning Kamp | UNIX since Zilog Zeus 3.20
phk at FreeBSD.ORG | TCP/IP since RFC 956
FreeBSD committer | BSD since 4.3-tahoe
Never attribute to malice what can adequately be explained by incompetence.

So thanks and so sorry for bothering you.
But my varnish is 1.1.1 not trunk, and I don't know how to fix at 1.1.1.
version 1.1.1 has no file related cache_backend_simple.c

Please give me a solution for the version 1.1.


Thanks for your help again.


>From: "Poul-Henning Kamp" <phk at phk.freebsd.dk>
>To: "Monty Ree" <chulmin2 at hotmail.com>
>CC: des at linpro.no, varnish-misc at projects.linpro.no
>Subject: Re: why purge makes child process restart?
>Date: Sun, 02 Sep 2007 07:38:02 +0000
>
>In message <BLU116-F54FAED96DBC408BD5A51E85CF0 at phx.gbl>, "Monty Ree"
writes:
> >Thanks for your help..
> >
> >Can you give us temporary vcl or urgent patch which drop the GET, HEAD
or
> >purge request which is not including Host header?
>
>This is what you need:
>
> http://varnish.projects.linpro.no/changeset/1930
>
>--
>Poul-Henning Kamp | UNIX since Zilog Zeus 3.20
>phk at FreeBSD.ORG | TCP/IP since RFC 956
>FreeBSD committer | BSD since 4.3-tahoe
>Never attribute to malice what can adequately be explained by incompetence.

_________________________________________________________________
??? ???? ?? 1G ?? ??!
http://im.msn.co.kr/new/function/function_02_11.asp

Monty Ree wrote:
> So thanks and so sorry for bothering you.
> But my varnish is 1.1.1 not trunk, and I don't know how to fix at 1.1.1.
> version 1.1.1 has no file related cache_backend_simple.c
>
> Please give me a solution for the version 1.1.
>

I imagine that you're best bet would be to run varnish from trunk
(http://varnish.projects.linpro.no/wiki/Installation).

-- james

"Monty Ree" <chulmin2 at hotmail.com> writes:
> Can you give us temporary vcl or urgent patch which drop the GET, HEAD
> or purge request which is not including Host header?

vcl_recv() {
if (!req.http.host) {
error 400 "No Host: header";
}
}

DES
--
Dag-Erling Sm?rgrav
Senior Software Developer
Linpro AS - www.linpro.no

james <james at nyi.net> writes:
> I imagine that you're best bet would be to run varnish from trunk
> (http://varnish.projects.linpro.no/wiki/Installation).

That might not be a good idea, considering the pace of development on
trunk.

DES
--
Dag-Erling Sm?rgrav
Senior Software Developer
Linpro AS - www.linpro.no

Hi!

Thanks for the bugreport and for the fix. No child restarts after the vcl fix
so far.

Checked out revision 1934 from the trunk but sending request with no Host
still crashes the slave process.
Adding
if (!req.http.host) {
? ? ? ? error 400 "No Host: header";
? ? }

works though.

Maybe that's because I chose backend based on the Host header:
sub vcl_recv {
if (req.http.host == "www.sdfsdf.lv") {
set req.backend = wwwa;
} elsif (req.http.host == "sdfhsdfh.lv") {
set req.backend = wwwb;
} elsif (req.http.host == "www.asdfsd.lv") {
...
} else {
error 404 "Unknown virtual host";
}

janis

On Sunday 02 September 2007 10:38, Poul-Henning Kamp wrote:
> In message <BLU116-F54FAED96DBC408BD5A51E85CF0 at phx.gbl>, "Monty Ree" writes:
> >Thanks for your help..
> >
> >Can you give us temporary vcl or urgent patch which drop the GET, HEAD or
> >purge request which is not including Host header?
>
> This is what you need:
>
> http://varnish.projects.linpro.no/changeset/1930