Quote:
Originally Posted by brian Actually the bulk of the upgrade time is usually spent in schema migrations and rebuilding the slapd indexes. These are almost exclusively IO bound processes. So faster and more disks almost always speed this up. |
That could be. I had top running in a virtual terminal during the process, and the times I looked over at it always showed a java process at >= 86% CPU. Not scientific in the least, just what I saw.
Quote:
Certainly a Intel P3 with software raid is going to be CPU bound, but this box is probably just fine for 12 users.
-bp
|
It's a temp box while we evaluate/test Zimbra. The server that will be its eventual home is a dual-Opteron box with 8 GB RAM and 12x 400 GB HD in RAID5. Zimbra will hopefully be running in a Xen VM on this box.