Update failure

[Bill Davidsen]

Patrick O'Callaghan wrote:
> On Mon, 2009-10-05 at 08:03 -0400, Fabio Jara wrote:
>>> I hope Tim (ignored_mailbox) will not see your advice. He recently
>> was
>>> very angry about yum clean all. Now it's forbidden to give such
>> advice
>>> ;-)
>> Why is that? that worked for me two times in a row now, why is it
>> forbidden? O_o For future reference.
> 
> The only reason to use "yum clean all" instead of "yum clean metadata"
> is if you're running out of space because of cached packages. Then you
> get to waste bandwidth (not to mention server resources) downloading
> them again.
> 
> "yum clean metadata" has solved every problem I've ever had that "yum
> clean all" would have solved. I would be interested to hear *reasoned*
> argument about why this might not always be the case (apart from the
> disk space issue already mentioned).
> 
I have a single /var/cache/yum filesystem on NFS which allows me to not only 
save disk and bandwidth, but to speed updates considerably. I'm sure using it 
from more than one machine at a time could cause a problem, but it is a useful 
stopgap here. It avoids having a mirror of everything, while pulling down each 
rpm only once, and only those actually used here.

The ideal solution would take care of this, but with only two admins doing 
updates manual sync works for us. I had a tool to do a more perfect job of 
contention control, but it proved to be overkill.

-- 
Bill Davidsen <davidsen at tmr.com>
   "We have more to fear from the bungling of the incompetent than from
the machinations of the wicked."  - from Slashdot