ke, 2007-10-31 kello 02:06:40 -0700, Jonathan Ryshpan kirjoitti: > On Tue, 2007-10-30 at 20:09 -0400, Tom Horsley wrote: > > I tried running an update, and I get a large number > > of these, followed by a message that there are no > > more mirrors to try: > > > > 20:07:54 : Trying other mirror. > > 20:07:57 : Failure getting http://mirror.stanford.edu/fedora/linux/updates/7/x86_64/repodata/comps-f7.xml: > > 20:07:57 : --> [Errno -1] Metadata file does not match checksum > > > > The i386 mirrors don't seem to have the same problem. > > I see had exactly the same problem for the last two days, running yumex. > The x86_64 repos appear to be corrupt while the i386 repos are OK. > > Could this have something to do with FC8, which should be out in a week > or so? > Googling a bit reveals that the same problem with comps-f7.xml has been in existence at least since early June, ie. since the release of F7. See for example: http://www.fedoraforum.org/forum/showthread.php?t=168232 It looks like update metadata is contained in comps-f7.xml - and GUI updaters such as pup and yumex use this file to fetch information on packages to be updated. If comps-f7.xml is corrupted (like in this case for x86_64 systems), pup and yumex visit every available mirror only to receive the same corrupt file from each and everyone of them - which in each case results in "incorrect checksum" error message. No wonder it takes ages to load the same metadata file over and over again from various mirrors and find all of them defective. Although updates with pup have been unsuccessful in this way a couple of times before, I haven't posted about it because yum has worked always when pup has failed. From this I conclude that yum doesn't use comps-f7.xml for anything. I can't help wondering, however, why should pup or yumex use .xml files? Isn't it a M$ standard - and controversial at that? Regards Antti
