On Fri, 2003-01-17 at 09:51, R P Herrold wrote: > or maybe a yum pid check at startup. ... > > I was out at a client with intermittent internet access -- I > found this on one of the servers. > > check-packages, and yum each are querying into the rpm > database. It is an almost certainty that there is a deadly > embrace on the database here. > > I'll reboot and analyze in a moment -- but at a minimum, only > one instance should be running at any time. A lockfile > mechanism for the cron version seems needed, if not the more > formal approach of checking for a valid yum pid lockfile at > startup, and bailing if found. > I think this should be fixed now - the locking I implemented in the most recent daily should do the right thing. -sv