On Thu, 2008-02-21 at 09:38 -0500, James Antill wrote: > On Wed, 2008-02-20 at 22:57 -0500, seth vidal wrote: > > Encountered a bug tonight that got me thinking about our locking and > > some issues. > > > > 1. unless an app locks, explicitly, nothing in our current > > infrastructure enforces it to happen. > > > > 2. if a user is running on their own cachedir from two different > > instances the instances will happily trample over each other and do > > 'odd' things > > > > 3. the same is true for misbehaving root-run apps, too. > > > > With this in mind would it be a good idea to change yum's lock mechanism > > such that: > > > > - the global lock means "the rpmdb is locked/transaction in progress" > > - each repo cachedir had a lockfile that could be placed to keep that > > repository dir from being changed > > So I think I agree with everyone when I say automatic locking would be > awesome, but... if we try to lock each repo cachedir separately , as we > use it, can't we get deadlock due to different instances locking repo > cache dirs in a different order?
We can get delayed but I don't think we can get deadlocked. -sv _______________________________________________ Yum-devel mailing list Yum-devel@linux.duke.edu https://lists.dulug.duke.edu/mailman/listinfo/yum-devel
