On Mon, 2008-04-14 at 13:13 -0400, Bill Nottingham wrote: > Callum Lerwick (seg haxxed com) said: > > > Is it a bug that nothing in the base system has a dependency on > > > event-compat-sysv? Relying on comps groups alone isn't entirely robust. > > > This seems to be also how some F8 to F9 upgrade failures are getting > > > stuck during boot as well. > > > > Seems like a bug to me. This breaks yum upgrades. > > Do you have your yum/upgrade log? There's no reason it shouldn't be > pulled in on upgrades. No, but I have 5 other machines to upgrade now that I know the problem... :) On another machine, if I do a full "yum upgrade", event-compat-sysv does indeed get pulled in. ...But that's not what I did. My laptop is rather limited in resources, "only" 512mb RAM and a couple GB free, so I figured it would be best to do a staged upgrade. First I decided to upgrade my key applications, "yum update galeon evolution". Through a long convoluted chain of dependencies, this ends up upgrading mkinitrd, which requires updating initscripts, which pulls in upstart. This resulted in a raft of file collisions with sysvinit, and fails during the transaction test. To get around it, I just "rpm -e --nodeps sysvinit" figuring upstart was supposed to replace it anyway. I assumed package dependencies would do their job and fix things or at the very least alert me to any problems. Obviously, this did not happen. After the first run, I did a full "yum update" to get the rest of the packages, this fails to bring in event-compat-sysv since I manually removed sysvinit, leaving you with an un-bootable system even though "package-cleanup --problems" indicates no problems.
Description: This is a digitally signed message part