Hi, Before I execute the openafs 1.4.1 to 1.4.5 upgrade I ran a sanity check of our OpenAFS deployment here. I inherited this system, I wasn't the original architect and I'm a new hire here. When I arrived a svn checkout of our source code to someone's afs home took hours. So I ran some tests and noticed file creating, renaming, and deletion took forever. Metadata issues. Queries to the VLDB took forever. Before I executed the updgrade I ran a sanity check, I was surprised at what I found! Only 1 vlserver, 1 ptserver, old entries of retired servers in vldb. So I deleted the bogus entries in VLDB. Migrated 2 home volumes off a desktop that was configured As a server to one of our enterprise class afs servers. Retired the desktop as a volume server. Added 2 vlservers, added two ptservers. I rebooted the Afs servers... noticed afs02 had a bogus uuid, deleted the sysid file, rebooted the server. All server processes came up. My control center script now execute in 5 seconds! ...My first day here I swear it took over 45 minutes! It was great to hear unsolicited "What did you do?", "Thanks"., "Great job". As it turns out the svn checkout to their afs home directories are fast! No more pain! I am working on installing izozone and some graphing utilities. Also..Bonnie results showed I went up from creating 23 files/second in AFS space to 653 files/second! And this is afs 1.4.1
So.. any of you out there that are experiencing AFS slowness, do a sanity check to see what you come up with. You might just say, WTF! _______________________________________________ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info