[users] Metadata problems
Dag Wieers
dag at wieers.com
Mon Dec 17 10:42:01 CET 2007
On Mon, 17 Dec 2007, Filipe Brandenburger wrote:
> On Dec 15, 2007 8:20 AM, Dag Wieers <dag at wieers.com> wrote:
>
> > The problem is that yum downloads the repomd.xml file from _one_ of the
> > mirrors. This file contains checksums for each repomd file (other.xml.gz,
> > filelist.xml.gz and primary.xml.gz)
> >
> > http://apt.sw.be/redhat/el5/en/i386/dag/repodata/repomd.xml
> >
> > The problem is that yum downloads this file from one of the mirrors in the
> > mirrorlist, and then compares the checksum with those other metadata files
> > it downloads from other mirrors.
>
> Ok, but then, why did it fail exactly when downloading the primary.xml.gzfrom
> apt.sw.be? If the repomd.xml supposedly came from the same source? From the
> output on David's e-mail, apt.sw.be was the first mirror tried:
Because the list of checksums was not from that server, I would think.
But maybe my analysis based on output from users is wrong. Please prove me
wrong and tell me how to fix it.
> I mean, I would admit these errors happening when the external mirrors are
> not in sync, but if apt.sw.be is not in sync with itself, then I guess the
> problem is not with yum...
If that is happening, sure. Proof it to me. Everybody is complaining, but
nobody is debugging and analysing. Apt and smart does not have this
problem using the same metadata (but no mirrors).
In fact, if you only use apt.sw.be, there is NO problem.
> Anyway, I also have these problems often with rpmforge, but I don't have
> them with CentOS repositories though, and CentOS has mirrors, and I'm pretty
> sure they don't control all their mirrors too.
But they do track their mirrors and modify the mirrorlist accordingly.
> Isn't it possible that the problem is on the script used for mirroring
> instead of on yum only?
If you come up with the cause, I can fix the problem. What I gave you was
my theory.
--
-- dag wieers, dag at wieers.com, http://dag.wieers.com/ --
[Any errors in spelling, tact or fact are transmission errors]
More information about the users
mailing list