On Mon, 2008-04-07 at 09:46 -0400, seth vidal wrote: > On Mon, 2008-04-07 at 16:36 +0300, Gilboa Davara wrote: > > On Mon, 2008-04-07 at 08:34 -0400, seth vidal wrote: > > > On Mon, 2008-04-07 at 11:00 +0300, Gilboa Davara wrote: > > > > Hello all, > > > > > > > > B9 updated to latest running on a VMWare virtual machine with 1GB RAM + > > > > 512MB swap. > > > > Running yum update dies on > > > > Updating: glib2 ... > > > > error: Couldn't fork %post: Cannot allocate memory. > > > > > > > > > > Okay so for x86_64 rawhide yum is going to use about 450-500MB of memory > > > for the package lists and depsolving. > > > > Eeeek. > > Are you expecting the same memory consumption under F9-release? > > For the same set of packages, etc on a 32bit box the memory size is > 112MB. Oh, that's far better. > > It's due to how python does 64bit objects in memory - they explode about > 2-3x the size. > > James Antill has done some marvelous work reducing the overall memory > footprint of yum, but it's not going to overcome the 64bit issue above. > > Unless there is a compelling reason like you're using > 4GB of ram, you > should think about running i686 not x86_64. Yeah, you're right. I use 64bit on the guests just for the sake of compatibility. Guess I'll have to double the memory on 64bit guests. P.S. I'd update the release notes about it: (Read: Use 32bit if you have less then 2GB of memory) > > > > > However, where you're seeing > > > things run out of memory is where rpmlib is running the transaction. > > If > > > you file this as a bug, file it against rpm since that's where the > > > memory use is increasing so much. > > > > Done. [1] > > Thank you. I've retained the current snapshot (packages in cache; yum update aborted) let me know if there's something that you want me to run before I finish the update process manually. (rpm -Uvh --force /var/cache/yum/rawhide/packages/*.rpm) - Gilboa -- fedora-test-list mailing list fedora-test-list@xxxxxxxxxx To unsubscribe: https://www.redhat.com/mailman/listinfo/fedora-test-list
