On Mon, 2004-06-14 at 21:09, Don Russell wrote: > So.... could the list of mirrors eligible for participating in up2date be > updated dynamically so as only synch'd mirrors would be used by up2date? > <SNIP> > It seems kind of sloppy, and a waste of bandwidth to try over and over > until it works... but maybe the effort to fix it is too high... > This is a interesting idea. The problem is how to create such list. Suppose that a cron job conects to each mirror server , downloads the list of *all* files , diffs them , do some parsing and then only echoes them to the mirror list sent to clients.. This is also a waste of bandwidth , considering that the script would have to be smart enough to only check the recently changed directories (or only the updates directory) , but at least it is a smaller waste than each user having to connect several times to find a mirror which has all the updates.. > Right now I have a red ball exclaming "There be updates! ... There be > updates!" ... It says 5 are available... when I launch up2date, I see only > 3... two of which appear to be mutually exclusive > (cdrecord/cdrecord-devel) > They're not mutually exclusive. They're complementary.. cdrecord-devel contains stuff needed to make programs based on cdrecord... And cdrecord contains the cdrecord itself... > Is the count mismatch due to different mirrors having different software > at different times? Once all the mirrors are in synch then up2date will > report things more accurately? Exactly. Each mirror takes a different time to sync with redhat. Some sync faster , some sync slower. As soon as all mirrors are in sync , up2date will show the right thing... -- Pedro Macedo