On Tue, 2013-06-18 at 15:20 +0200, Zdenek Pavlas wrote: > On a filesystem with 1s mtime precision a RPM transaction ran > immediately after Yum may alter rpmdb but keep the timestamp.
NAK. Not even sure it's worth trying to workaround this, if we do though I'd guess the "best" way of doing it would be to alter the mtime of /var/lib/rpm/Packages and /var/lib/yum/../version to be the same time ~1 second in the past (when we write out the /version file). But that feels pretty horrible. A better workaround IMO would be to put a 1 second sleep in the rpm command. Not sure Panu would go for that though :). But, again, given that they need: 1. To be not using an old FS without good timestamp precision. 2. To use rpm directly, which we complain about anyway. 3. To run an rpm transaction within the same second that the last yum transaction finished. ...I'm leaning more towards NaB, stop doing that. The below patch breaks the common case, making the generation of the "/version" cache mostly useless (it's always generated within the same second that we last altered the rpmdb AFAIK). _______________________________________________ Yum-devel mailing list [email protected] http://lists.baseurl.org/mailman/listinfo/yum-devel
