On Sat, 2006-06-10 at 14:19 +1000, David Timms wrote:
> Jesse Keating wrote:
> > On Fri, 2006-06-09 at 22:00 +1000, David Timms wrote:
> >> Note all the stuff in the actual header, especially:
> >> HTTP/1.1 404 Not Found
> >> Throwing away the header that has the correct info in it is what is
> >> going wrong. From what someone said urlgrabber is used in yum ->
> >> shouldn't urlgrabber then return fail if the server is stating html
> >> errors eg 404, no matter how pretty the 404 looks to a human using
> >> their
> >> web browser ?
> >
> > Didn't Seth say that f.r.c was fixed and thus would not make a very good
> > test case?
> Agreed. That would be why I included the problem {pacific internet} from
> Sean's report. Also the mentioned ~"302 and redirect" is another html
> response ie not 200. So could someone list the settings that would have
> caused the 200+text404 response, so that it could be used as an
> additional test case ?
>
> Still, it would be great to solve the garbage in {=dodgy repodata} >
> getting stuck problem. Looking for well formed xml in the data appears
> to be a good place to start, should be seeing:
> <?xml version="1.0" encoding="UTF-8"?>
> <repomd xmlns="http://linux.duke.edu/metadata/repo";> ...
It doesn't fail. I tested 2.6.X and cvs today.
When yum encounters a broken repomd.xml it marks the mirror as bad and
goes to the next one.
If you only have one mirror it exits.
go ahead, try it.
make a repository
set it up in your yum configuration
then copy some random html file over the repomd.xml
then setup a second baseurl for the repository that actually works.
then run yum list available
watch as the broken repomd.xml gets read, displays some crap from the
library about the broken format then moves on to the next baseurl for
that repo.
seriously, try it. :)
-sv
--
fedora-devel-list mailing list
fedora-devel-list@xxxxxxxxxx
https://www.redhat.com/mailman/listinfo/fedora-devel-list
