ESI include inserts garbage code
Hi folks,
I'm experimenting with ESI and something weird has come up: varnish is
adding garbage code to the output.
The public address is the production varnish that doesn't use this yet,
so you'll need to edit your hosts-file:
81.23.226.44 www.mobilecowboys.nl
My sample page is http://www.mobilecowboys.nl/news/hbptest
It switches randomly between 3 diferent includes all of which are cached:
http://www.mobilecowboys.nl/news/hbpplayer/review_id/1370977981
http://www.mobilecowboys.nl/news/hbpplayer/review_id/1600813923
http://www.mobilecowboys.nl/news/hbpplayer/review_id/1271683253
The extra code looks like checksums. They are consistent for each include.
The relevant debug lines from varnishlog are:
7 Debugc "tag {include
src="/news/hbpplayer/review_id/1370977981"} 0 1 0"
7 Debugc "Incl " src="/news/hbpplayer/review_id/1370977981"""
7 Debugc "tag {include
src="/news/hbpplayer/review_id/1600813923"} 0 1 0"
7 Debugc "Incl " src="/news/hbpplayer/review_id/1600813923"""
7 Debugc "tag {include
src="/news/hbpplayer/review_id/1271683253"} 0 1 0"
7 Debugc "Incl " src="/news/hbpplayer/review_id/1271683253"""
And near the end of a request:
7 Debugc "herding"
I'm running varnish 2.0.6 from ports on freebsd 7.2
Thanks,
Bert-Jan
___
varnish-misc mailing list
varnish-misc@projects.linpro.no
http://projects.linpro.no/mailman/listinfo/varnish-misc
Re: ESI include inserts garbage code
In message <4b748e05.3060...@bugbyte.nl>, Bert-Jan de Lange writes: >Hi folks, > >I'm experimenting with ESI and something weird has come up: varnish is >adding garbage code to the output. What is the garbage that you are seing ? If it a separate line with hex digits, then it is the chunked encoding we use to transmit ESI with to HTTP/1.1 clients... -- Poul-Henning Kamp | UNIX since Zilog Zeus 3.20 p...@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. ___ varnish-misc mailing list varnish-misc@projects.linpro.no http://projects.linpro.no/mailman/listinfo/varnish-misc
Re: ESI include inserts garbage code
On 12-Feb-2010 00:28, Poul-Henning Kamp wrote: > In message<4b748e05.3060...@bugbyte.nl>, Bert-Jan de Lange writes: > >> Hi folks, >> >> I'm experimenting with ESI and something weird has come up: varnish is >> adding garbage code to the output. >> > What is the garbage that you are seing ? > > If it a separate line with hex digits, then it is the chunked encoding > we use to transmit ESI with to HTTP/1.1 clients... > Yes, exactly that. It looks like this: 1a 53c Blij verrast waren we toen we de Nokia 5630 Xpressmusic uit de doos haalde. Op het eerste gezicht ziet het toestel van Nokia er namelijk erg goed uit. Als je daarnaast op basis van de naam ook een goede mp3-speler mag verwachten, stijgt de spanning helemaal. Maar zit er ook echt muziek in deze telefoon? Lees de volledige review van de Nokia 5630 op BesteProduct.nl 19 0 ___ varnish-misc mailing list varnish-misc@projects.linpro.no http://projects.linpro.no/mailman/listinfo/varnish-misc
Re: ESI include inserts garbage code
In message <4b749359.3080...@bugbyte.nl>, Bert-Jan de Lange writes: >On 12-Feb-2010 00:28, Poul-Henning Kamp wrote: >> If it a separate line with hex digits, then it is the chunked encoding >> we use to transmit ESI with to HTTP/1.1 clients... >> >Yes, exactly that. It looks like this: Your client should not show you that. I fixed a bug in -trunk three days that might relate to this, (HTTP/1.0 vs HTTP/1.1 confusion). Or you can try in vcl_deliver to always force the reply protocol to be HTTP/1.1 -- Poul-Henning Kamp | UNIX since Zilog Zeus 3.20 p...@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. ___ varnish-misc mailing list varnish-misc@projects.linpro.no http://projects.linpro.no/mailman/listinfo/varnish-misc
Re: ESI include inserts garbage code
On 12-Feb-2010 00:48, Poul-Henning Kamp wrote: > In message<4b749359.3080...@bugbyte.nl>, Bert-Jan de Lange writes: > >> On 12-Feb-2010 00:28, Poul-Henning Kamp wrote: >> > >>> If it a separate line with hex digits, then it is the chunked encoding >>> we use to transmit ESI with to HTTP/1.1 clients... >>> >>> >> Yes, exactly that. It looks like this: >> > Your client should not show you that. > > I fixed a bug in -trunk three days that might relate to this, > (HTTP/1.0 vs HTTP/1.1 confusion). > > Or you can try in vcl_deliver to always force the reply protocol > to be HTTP/1.1 > Yes, I see it responds with HTTP/1.0. I'll fix it in vcl_deliver, no problem. Thanks ! ___ varnish-misc mailing list varnish-misc@projects.linpro.no http://projects.linpro.no/mailman/listinfo/varnish-misc
Re: ESI include inserts garbage code {SOLVED]
On 12-Feb-2010 00:48, Poul-Henning Kamp wrote:
> In message<4b749359.3080...@bugbyte.nl>, Bert-Jan de Lange writes:
>
>> On 12-Feb-2010 00:28, Poul-Henning Kamp wrote:
>>
>
>>> If it a separate line with hex digits, then it is the chunked encoding
>>> we use to transmit ESI with to HTTP/1.1 clients...
>>>
>>>
>> Yes, exactly that. It looks like this:
>>
> Your client should not show you that.
>
> I fixed a bug in -trunk three days that might relate to this,
> (HTTP/1.0 vs HTTP/1.1 confusion).
>
> Or you can try in vcl_deliver to always force the reply protocol
> to be HTTP/1.1
>
>
Added to the end of vcl_deliver:
if (resp.http.Transfer-Encoding == "chunked") {
set resp.proto = regsub(resp.proto, "1\.0", "1.1");
}
The checksums are gone now.
___
varnish-misc mailing list
varnish-misc@projects.linpro.no
http://projects.linpro.no/mailman/listinfo/varnish-misc
Re: ESI include inserts garbage code {SOLVED]
In message <4b749cb9.6080...@bugbyte.nl>, Bert-Jan de Lange writes: >The checksums are gone now. Well, they're not really checksums, they are byte counts, but... -- Poul-Henning Kamp | UNIX since Zilog Zeus 3.20 p...@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. ___ varnish-misc mailing list varnish-misc@projects.linpro.no http://projects.linpro.no/mailman/listinfo/varnish-misc
