From fabrice at dulaunoy.com  Wed Jul 18 11:38:23 2007
From: fabrice at dulaunoy.com (fabrice at dulaunoy.com)
Date: Wed, 18 Jul 2007 13:38:23 +0200 (CEST)
Subject: Error xxx Guru Meditation
Message-ID: <Pine.LNX.4.58.0707181223580.17210@casca.cegeka.be>

Dear all,

I have some customer which are complaining about the "design" of the
error page generated when all backend server are down.

To fix this aesthetical request , I've made a small patch.
This patch allow to provide a folder ( by the use of the flag -e
/path/to/error_/folder ) where html error page belong.
If you provide a 503.html page in this folder, you've got that page in
place of the default "Guru meditation" message.

May be this patch could be useful for someone else.

Best regards and thanks a lot for the varnish product.


DULAUNOY Fabrice <fabrice at dulaunoy.com>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: varnish-1.0.4.diff
Type: text/x-diff
Size: 13939 bytes
Desc: 
URL: <https://www.varnish-cache.org/lists/pipermail/vinyl-dev/attachments/20070718/bd6fb382/attachment.bin>

From rudd-o at rudd-o.com  Thu Jul 19 14:52:31 2007
From: rudd-o at rudd-o.com (Manuel Amador (Rudd-O))
Date: Thu, 19 Jul 2007 09:52:31 -0500
Subject: Error xxx Guru Meditation
In-Reply-To: <Pine.LNX.4.58.0707181223580.17210@casca.cegeka.be>
References: <Pine.LNX.4.58.0707181223580.17210@casca.cegeka.be>
Message-ID: <1184856751.6380.4.camel@karen>

I know I'm not exactly persona grata, but I would love for this patch to
be considered as well.  Just a quick serveup of an HTML file for the
client that can be customized by the user.

On mi?, 2007-07-18 at 13:38 +0200, fabrice at dulaunoy.com wrote:
> Dear all,
> 
> I have some customer which are complaining about the "design" of the
> error page generated when all backend server are down.
> 
> To fix this aesthetical request , I've made a small patch.
> This patch allow to provide a folder ( by the use of the flag -e
> /path/to/error_/folder ) where html error page belong.
> If you provide a 503.html page in this folder, you've got that page in
> place of the default "Guru meditation" message.
> 
> May be this patch could be useful for someone else.
> 
> Best regards and thanks a lot for the varnish product.
> 
> 
> DULAUNOY Fabrice <fabrice at dulaunoy.com>
> _______________________________________________ varnish-dev mailing list varnish-dev at projects.linpro.no http://projects.linpro.no/mailman/listinfo/varnish-dev
	Manuel Amador (Rudd-O) <rudd-o at rudd-o.com>
	The R Zone - http://rudd-o.com/
	GPG key ID 0xC8D28B92 at http://wwwkeys.pgp.net/

Now playing, courtesy of Amarok: 
You will be honored for contributing your time and skill to a worthy cause.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 189 bytes
Desc: This is a digitally signed message part
URL: <https://www.varnish-cache.org/lists/pipermail/vinyl-dev/attachments/20070719/fe7d97e0/attachment.bin>

From andrearo at pvv.ntnu.no  Thu Jul 19 15:13:32 2007
From: andrearo at pvv.ntnu.no (=?ISO-8859-1?Q?Andreas_R=F8sdal?=)
Date: Thu, 19 Jul 2007 17:13:32 +0200 (CEST)
Subject: Error xxx Guru Meditation
In-Reply-To: <1184856751.6380.4.camel@karen>
References: <Pine.LNX.4.58.0707181223580.17210@casca.cegeka.be>
	<1184856751.6380.4.camel@karen>
Message-ID: <20070719170659.F52658@verden.pvv.ntnu.no>

> On mi?, 2007-07-18 at 13:38 +0200, fabrice at dulaunoy.com wrote:
>> Dear all,
>>
>> I have some customer which are complaining about the "design" of the
>> error page generated when all backend server are down.
>>
>> To fix this aesthetical request , I've made a small patch.
>> This patch allow to provide a folder ( by the use of the flag -e
>> /path/to/error_/folder ) where html error page belong.
>> If you provide a 503.html page in this folder, you've got that page in
>> place of the default "Guru meditation" message.
>>
>> May be this patch could be useful for someone else.

Ideally, when all backend servers are down, varnish should show an old 
cached version of the requested resource if possible. It would mean that 
backend failures wouldn't be noticed by end users. Would this be 
possible, or planned ?

  - Andreas

From rudd-o at rudd-o.com  Thu Jul 19 15:27:18 2007
From: rudd-o at rudd-o.com (Manuel Amador (Rudd-O))
Date: Thu, 19 Jul 2007 10:27:18 -0500
Subject: Error xxx Guru Meditation
In-Reply-To: <20070719170659.F52658@verden.pvv.ntnu.no>
References: <Pine.LNX.4.58.0707181223580.17210@casca.cegeka.be>
	<1184856751.6380.4.camel@karen>
	<20070719170659.F52658@verden.pvv.ntnu.no>
Message-ID: <1184858838.6380.13.camel@karen>

On jue, 2007-07-19 at 17:13 +0200, Andreas R?sdal wrote:

> Ideally, when all backend servers are down, varnish should show an old 
> cached version of the requested resource if possible. It would mean that 
> backend failures wouldn't be noticed by end users. Would this be 
> possible, or planned ?

This would also be a good idea but I'd hate to see this be the
"default", because it can mask *serious* HTTP backend errors, misleading
administrators into believing the site is running OK.  At least, it
should never trigger if a "Pragma: no-cache" header is sent by the
client, so that the ones among us using Wget/munin/nagios to monitor our
sites can actually monitor them :-).

> 
>   - Andreas
	Manuel Amador (Rudd-O) <rudd-o at rudd-o.com>
	The R Zone - http://rudd-o.com/
	GPG key ID 0xC8D28B92 at http://wwwkeys.pgp.net/

Now playing, courtesy of Amarok: Andrea - Time to pray
What good is an obscenity trial except to popularize literature?
		-- Nero Wolfe, "The League of Frightened Men"
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 189 bytes
Desc: This is a digitally signed message part
URL: <https://www.varnish-cache.org/lists/pipermail/vinyl-dev/attachments/20070719/01d86720/attachment.bin>

From darix at web.de  Mon Jul 23 13:52:03 2007
From: darix at web.de (Marcus Rueckert)
Date: Mon, 23 Jul 2007 15:52:03 +0200
Subject: Assert for error handling sucks
Message-ID: <20070723135203.GA26939@pixel.global-banlist.de>

hi,

as the subject says. While i was testing my 1.1 rpms for suse I ran into
a "crash" with varnish. I wondered as the error message want really
helpful.

[[[
$ /usr/sbin/varnishd -f /etc/varnish/vcl.conf -T127.0.0.1:6082 -s file,/var/cache/varnish/
file /var/cache/varnish//varnish.fTZLF4 (unlinked) size 46919348723712
bytes (11454919122 fs-blocks, 11454919122 pages)
Assert error in smf_initfile(), storage_file.c line 239:
  Condition((ftruncate(sc->fd, sc->filesize)) == 0) not true.
    errno = 27 (File too large)
]]]

Can this not be handled more gracefully?

For testing i did:
[[[
$/usr/sbin/varnishd -f /etc/varnish/vcl.conf -T127.0.0.1:6082 -s file,/var/cache/varnish/,1024 
Error: size too small, at least 524288 needed
]]]

now i wonder if 524288 is a good default value?
what would you suggest for a package default config?

with kind regards,

    darix

-- 
           openSUSE - SUSE Linux is my linux
               openSUSE is good for you
                   www.opensuse.org
-------------- next part --------------
(gdb) r
Starting program: /usr/sbin/varnishd -f /etc/varnish/vcl.conf -T127.0.0.1:6082 -s file,/var/cache/varnish
[Thread debugging using libthread_db enabled]
[New Thread 47252503396816 (LWP 25764)]
Error while mapping shared library sections:
./bin.XX8mRPjT: No such file or directory.
Error while reading shared library symbols:
./bin.XX8mRPjT: No such file or directory.
Error while reading shared library symbols:
./bin.XX8mRPjT: No such file or directory.
file /var/cache/varnish/varnish.8wVIfz (unlinked) size 47252501147648 bytes (11536255163 fs-blocks, 11536255163 pages)
Assert error in smf_initfile(), storage_file.c line 239:
  Condition((ftruncate(sc->fd, sc->filesize)) == 0) not true.
  errno = 27 (File too large)

Program received signal SIGABRT, Aborted.
[Switching to Thread 47252503396816 (LWP 25764)]
0x00002af9d48deaa5 in raise () from /lib64/libc.so.6
(gdb) bt
#0  0x00002af9d48deaa5 in raise () from /lib64/libc.so.6
#1  0x00002af9d48dfe60 in abort () from /lib64/libc.so.6
#2  0x00002af9d43717fa in lbv_assert (func=0x4303c3 "smf_initfile", file=0x4301b7 "storage_file.c", line=239, cond=0x4303d0 "(ftruncate(sc->fd, sc->filesize)) == 0", err=27)
    at assert.c:58
#3  0x0000000000424f2c in smf_initfile (sc=0x53d790, size=0x53d0f3 "", newfile=1) at storage_file.c:239
#4  0x0000000000425414 in smf_init (parent=0x53cf80, spec=0x7fffd685659f "/var/cache/varnish") at storage_file.c:334
#5  0x000000000042789a in setup_storage (s_arg=0x7fffd685659a "file,/var/cache/varnish") at varnishd.c:158
#6  0x0000000000428dba in main (argc=0, argv=0x7fffd6854aa8) at varnishd.c:573
(gdb) bt full
#0  0x00002af9d48deaa5 in raise () from /lib64/libc.so.6
No symbol table info available.
#1  0x00002af9d48dfe60 in abort () from /lib64/libc.so.6
No symbol table info available.
#2  0x00002af9d43717fa in lbv_assert (func=0x4303c3 "smf_initfile", file=0x4301b7 "storage_file.c", line=239, cond=0x4303d0 "(ftruncate(sc->fd, sc->filesize)) == 0", err=27)
    at assert.c:58
No locals.
#3  0x0000000000424f2c in smf_initfile (sc=0x53d790, size=0x53d0f3 "", newfile=1) at storage_file.c:239
        __func__ = "smf_initfile"
#4  0x0000000000425414 in smf_init (parent=0x53cf80, spec=0x7fffd685659f "/var/cache/varnish") at storage_file.c:334
        size = 0x53d0f3 ""
        p = 0x53d0e0 "/var/cache/varnish"
        q = 0x53d2d0 ""
        st = {st_dev = 2304, st_ino = 9979057, st_nlink = 3, st_mode = 16872, st_uid = 134, st_gid = 129, pad0 = 0, st_rdev = 0, st_size = 4096, st_blksize = 4096, st_blocks = 8,
  st_atim = {tv_sec = 1185132340, tv_nsec = 0}, st_mtim = {tv_sec = 1185132397, tv_nsec = 0}, st_ctim = {tv_sec = 1185132397, tv_nsec = 0}, __unused = {0, 0, 0}}
        sc = (struct smf_sc *) 0x53d790
        u = 33
        __func__ = "smf_init"
#5  0x000000000042789a in setup_storage (s_arg=0x7fffd685659a "file,/var/cache/varnish") at varnishd.c:158
        p = 0x7fffd685659e ",/var/cache/varnish"
        q = 0x7fffd685659f "/var/cache/varnish"
        stp = (struct stevedore *) 0x536920
        __func__ = "setup_storage"
#6  0x0000000000428dba in main (argc=0, argv=0x7fffd6854aa8) at varnishd.c:573
        o = -1
        C_flag = 0
        d_flag = 0
        F_flag = 0
        b_arg = 0x0
        f_arg = 0x7fffd6856570 "/etc/varnish/vcl.conf"
        f_fd = 6
        h_arg = 0x430f41 "classic"
        n_arg = 0x0
        P_arg = 0x0
        s_arg = 0x7fffd685659a "file,/var/cache/varnish"
        T_arg = 0x7fffd6856588 "127.0.0.1:6082"
        p = 0x2af900000001 <Address 0x2af900000001 out of bounds>
        cli = {{sb = 0x53a010, result = CLIS_OK}}
        pfh = (struct pidfh *) 0x0
        dirname = "/var/cache//varnish/zen\000?Gi??*\000\000pF\205??\177\000\000?Di??*\000\000?y??*\000\000s?%??*\000\001??\213??*\000\000 at Ii??*\000\000?F\205?\000\000\000\000(?6??*\000\000\000 %??*\000\000(?6??*\000\000\a\000\000\000\000\000\000\000&?\226??*\000\000?{\213??*\000\000 at Ii??*\000\000\000\000\000\000\000\000\000\000?\021\021\000\000\000\000\000?H\205??\177\000\000?J\205??\177\000\000xJ\205??\177\000\000\006\000\000\000\000\000\000\000\000\220y??*\000\000"...
        __func__ = "main"


#2  0x00002ac263c957fa in lbv_assert (func=0x4303c3 "smf_initfile", file=0x4301b7 "storage_file.c", line=239, cond=0x4303d0 "(ftruncate(sc->fd, sc->filesize)) == 0", err=27)
    at assert.c:58
#3  0x0000000000424f2c in smf_initfile (sc=0x53d790, size=0x53d2ec "", newfile=1) at storage_file.c:239
#4  0x0000000000425165 in smf_init (parent=0x53cf80, spec=0x7fff46f32596 "/var/cache/varnish/somefile") at storage_file.c:280
#5  0x000000000042789a in setup_storage (s_arg=0x7fff46f32591 "file,/var/cache/varnish/somefile") at varnishd.c:158
#6  0x0000000000428dba in main (argc=0, argv=0x7fff46f31178) at varnishd.c:573


From santiago.del.castillo at fnbox.com  Mon Jul 23 18:31:28 2007
From: santiago.del.castillo at fnbox.com (Santiago Del Castillo)
Date: Mon, 23 Jul 2007 15:31:28 -0300
Subject: New with varnis
Message-ID: <33d607980707231131ib1fe195lab7424e0433ca2e8@mail.gmail.com>

Hi, I'm willing to try varnish on our production environment. Right
now I've 5 squids running and serving static files. (1.3 million files
~40k each and counting)

I've some questions about varnish and the difference with squid.

- Is it possible to set the amount of RAM varnish should use?
- Varnish creates a directory tree like the one squid does?
- Varnish starts dumping content to disk once memory is full?
- '-s' option limits the disk size varnish will use to store files?


Nothing else to say (yet) except that varnish seems to be a very good
candidate for squid replacement.

Cheers!
Santiago


From phk at phk.freebsd.dk  Tue Jul 24 07:50:10 2007
From: phk at phk.freebsd.dk (Poul-Henning Kamp)
Date: Tue, 24 Jul 2007 07:50:10 +0000
Subject: New with varnis 
In-Reply-To: Your message of "Mon, 23 Jul 2007 15:31:28 -0300."
	<33d607980707231131ib1fe195lab7424e0433ca2e8@mail.gmail.com> 
Message-ID: <21946.1185263410@critter.freebsd.dk>

In message <33d607980707231131ib1fe195lab7424e0433ca2e8 at mail.gmail.com>, "Santi
ago Del Castillo" writes:
>Hi, I'm willing to try varnish on our production environment. Right
>now I've 5 squids running and serving static files. (1.3 million files
>~40k each and counting)
>
>I've some questions about varnish and the difference with squid.
>
>- Is it possible to set the amount of RAM varnish should use?

No, Varnish will use whatever is available because Varnish relies
on the Virtual Memory system of the kernel to do the RAM/disk
scheduling.

>- Varnish creates a directory tree like the one squid does?

No, just a single file.

>- Varnish starts dumping content to disk once memory is full?

Depends on the VM strategy of your operating system.  In general
you shouldn't need to worry about this at all.

>- '-s' option limits the disk size varnish will use to store files?

Yes.


-- 
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.


From rudd-o at rudd-o.com  Tue Jul 24 08:17:19 2007
From: rudd-o at rudd-o.com (Manuel Amador (Rudd-O))
Date: Tue, 24 Jul 2007 03:17:19 -0500
Subject: New with varnis
In-Reply-To: <21946.1185263410@critter.freebsd.dk>
References: <21946.1185263410@critter.freebsd.dk>
Message-ID: <1185265039.23239.5.camel@karen>

IF I understood Varnish's principle of operation, is that it mmaps a
large file -- in effect creating a sort of "RAM backed by disk".  Things
are flushed periodically, and whenever the computer is low on RAM
(memory pressure increases), it starts to remove things from RAM and
re-read things from disk as necessary.

Certainly makes more sense than the Squid approach.

On mar, 2007-07-24 at 07:50 +0000, Poul-Henning Kamp wrote:
> In message <33d607980707231131ib1fe195lab7424e0433ca2e8 at mail.gmail.com>, "Santi
> ago Del Castillo" writes:
> >Hi, I'm willing to try varnish on our production environment. Right
> >now I've 5 squids running and serving static files. (1.3 million files
> >~40k each and counting)
> >
> >I've some questions about varnish and the difference with squid.
> >
> >- Is it possible to set the amount of RAM varnish should use?
> 
> No, Varnish will use whatever is available because Varnish relies
> on the Virtual Memory system of the kernel to do the RAM/disk
> scheduling.
> 
> >- Varnish creates a directory tree like the one squid does?
> 
> No, just a single file.
> 
> >- Varnish starts dumping content to disk once memory is full?
> 
> Depends on the VM strategy of your operating system.  In general
> you shouldn't need to worry about this at all.
> 
> >- '-s' option limits the disk size varnish will use to store files?
> 
> Yes.
> 
> 
	Manuel Amador (Rudd-O) <rudd-o at rudd-o.com>
	The R Zone - http://rudd-o.com/
	GPG key ID 0xC8D28B92 at http://wwwkeys.pgp.net/

Now playing, courtesy of Amarok: You are scrupulously honest, frank, and straightforward.  Therefore you
have few friends.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 189 bytes
Desc: This is a digitally signed message part
URL: <https://www.varnish-cache.org/lists/pipermail/vinyl-dev/attachments/20070724/4e4136c9/attachment.bin>

From des at linpro.no  Tue Jul 24 09:24:42 2007
From: des at linpro.no (=?utf-8?Q?Dag-Erling_Sm=C3=B8rgrav?=)
Date: Tue, 24 Jul 2007 11:24:42 +0200
Subject: Assert for error handling sucks
In-Reply-To: <20070723135203.GA26939@pixel.global-banlist.de> (Marcus
	Rueckert's message of "Mon\, 23 Jul 2007 15\:52\:03 +0200")
References: <20070723135203.GA26939@pixel.global-banlist.de>
Message-ID: <87ejiy6tnp.fsf@des.linpro.no>

Marcus Rueckert <darix at web.de> writes:
> Can this not be handled more gracefully?

Sure it can, please send patches :)

> now i wonder if 524288 is a good default value?

It is an arbitrary minimum.  I can't imagine a setup where it would make
sense to run with that little cache, though.

> what would you suggest for a package default config?

I suggest reading the release engineering documentation in the wiki, and
subscribing to varnish-dist at projects.linpro.no as you were invited to do
a couple of months ago.

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


From andrearo at pvv.ntnu.no  Tue Jul 31 14:37:37 2007
From: andrearo at pvv.ntnu.no (=?ISO-8859-1?Q?Andreas_R=F8sdal?=)
Date: Tue, 31 Jul 2007 16:37:37 +0200 (CEST)
Subject: Varnish crashes, and Cache child dies
Message-ID: <20070731161701.V3904@verden.pvv.ntnu.no>

Hello!

I was really looking forward to varnish 1.1... We've just upgraded from 
varnish 1.0.4 to the new 1.1, and now varnish regularily crashes every 60 
seconds.

The VCL file is available here:
http://www.pvv.ntnu.no/~andrearo/vcl.conf

The operating system is SuSE Linux 9 x86_64. This is grandis01, for those 
who have seen this system before..  :) To me it looks like an assertion is 
triggered in cache_http.c.

Here is the output from running varnish in debug mode:

/usr/sbin/varnishd -f /etc/varnish/vcl.conf -T127.0.0.1:6082 -s 
file,/var/cache/varnish -d
file /var/cache/varnish/varnish.hswcHe (unlinked) size 5476352 bytes (1337 
fs-blocks, 1337 pages)
Using old SHMFILE
New Pid 10330

rolling(1)...
rolling(2)...
start
CLI <start>
start child pid 10351
200 0

Child said (2, 10351): <<Child starts
managed to mmap 5476352 bytes of 5476352
Ready
CLI ready
>>
Child said (2, 10351): <<socktest: linger=0 sndtimeo=0 rcvtimeo=0
>>
Child said (2, 10351): <<Assert error in http_Write(), cache_http.c line 
977:
   Condition((hp->hd[HTTP_HDR_STATUS].b) != 0) not true.
   errno = 0 (Success)
>>
Cache child died pid=10351 status=0x6
Clean child
Child cleaned
start child pid 10365
Child said (2, 10365): <<Child starts
managed to mmap 5476352 bytes of 5476352
Ready
CLI ready




  - Andreas


From des at linpro.no  Tue Jul 31 16:12:20 2007
From: des at linpro.no (=?utf-8?Q?Dag-Erling_Sm=C3=B8rgrav?=)
Date: Tue, 31 Jul 2007 18:12:20 +0200
Subject: Varnish crashes, and Cache child dies
In-Reply-To: <20070731161701.V3904@verden.pvv.ntnu.no> ("Andreas
	=?utf-8?Q?R=C3=B8sdal=22's?=
	message of "Tue\, 31 Jul 2007 16\:37\:37 +0200 \(CEST\)")
References: <20070731161701.V3904@verden.pvv.ntnu.no>
Message-ID: <874pjkwo0r.fsf@des.linpro.no>

Andreas R?sdal <andrearo at pvv.ntnu.no> writes:
>>>
> Child said (2, 10351): <<Assert error in http_Write(), cache_http.c line 
> 977:
>    Condition((hp->hd[HTTP_HDR_STATUS].b) != 0) not true.
>    errno = 0 (Success)
>>>

This is a known bug, see http://varnish.projects.linpro.no/ticket/128
for an explanation and a patch.

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


From andrearo at pvv.ntnu.no  Tue Jul 31 16:50:28 2007
From: andrearo at pvv.ntnu.no (=?ISO-8859-1?Q?Andreas_R=F8sdal?=)
Date: Tue, 31 Jul 2007 18:50:28 +0200 (CEST)
Subject: Varnish crashes, and Cache child dies
In-Reply-To: <874pjkwo0r.fsf@des.linpro.no>
References: <20070731161701.V3904@verden.pvv.ntnu.no>
	<874pjkwo0r.fsf@des.linpro.no>
Message-ID: <20070731184821.V5166@verden.pvv.ntnu.no>

On Tue, 31 Jul 2007, [utf-8] Dag-Erling Sm?rgrav wrote:
> Andreas R?sdal <andrearo at pvv.ntnu.no> writes:
>>>>
>> Child said (2, 10351): <<Assert error in http_Write(), cache_http.c line
>> 977:
>>    Condition((hp->hd[HTTP_HDR_STATUS].b) != 0) not true.
>>    errno = 0 (Success)
>>>>
>
> This is a known bug, see http://varnish.projects.linpro.no/ticket/128
> for an explanation and a patch.
>
> DES


Thanks, I'll be trying out Varnish 1.1.1 once that is released then.

  - Andreas