On Tue, 2004-06-15 at 17:47, seth vidal wrote: > On Tue, 2004-06-15 at 17:09 +0530, Sanjay Arora wrote: > > Now the scenario is this. I have a local yum-repository of RH9 updates, > > from which I deleted several corrupted rpms, so it is missing some > > rpms...about 15 of them. > > > > In the update url, I have given two urls...first of the local repository > > and second of the duke mirror. Now, when yum reaches the first missing > > file, it switches to the second failover url of the duke mirror, but it > > does not switch back to the faster and local mirror for getting other > > files. It keeps downloading files from the remote, ignoring the local > > repository. > > > > That's correct, once a mirror has failed it is never restored. Why would > yum oscillate back and forth if it has already experienced one as a > failed mirror? Well Seth It does not matter when you have got a fat pipe and can connect to multiple repositories. But, believe me...if you got a slow dial-up link and yum can recognize a LOCAL repository (on the Lan), through a configuration parameter, and prefers it over a remote repository and uses the remote one only for the failed files...it really makes a difference. Of course, now the burden to ensure that local repository files are uncorrupted, falls upon the user. Esentially one will have to write a script to checksum & verify the file, delete & re-download the files that fail. If these features are built into yum, it will help the new users a lot. Rsync can obviate the need to do that but it is not at all suited to dial-up links prone to disconnections. IAC, Tell me a bit about yourself.You seem to reply to most of the mails on this list ;-) Best regards. Sanjay.
