Hi, --On 7. November 2006 20:05:51 -0800 David Severance <[EMAIL PROTECTED]> wrote:
You see, I recognize we have reached the limits of what I can do with NFS. I'd like to offer my users more but it seems like in order to do that I have to accept a single point of failure model. I had thought we were quite past the point in time where that was considered an acceptable model in a large enterprise type environment even if that is a University. Perhaps I need to make things even more complex and use some sort of VMware scheme to have virtual imap servers that can fail over for each other but again it's another complex unproven approach unless someone on this list has done that, please feel free to speak up. Does everyone just accept this single point of failure issue without concern? I'm finding it a hard one to sallow that with 40K users to support.
this isn't with UW any more, because we switched to Cyrus a few years ago, but we have a High Availability Cluster with RHEL 3's Cluster Suite. The mail files reside on a SAN (with redundant connections from each node that use separate switches). Only one node is active as IMAP server at any given moment, but when one goes down (for maintenance or due to a crash) the other node takes over. Of course IMAP sessions will be interrupted when that happens, but that's acceptable to us. Note that we *don't* use a cluster file system but standard ext3. That means that the partitions are mounted and unmounted as needed, because only one node may mount any given partition at a time.
Initially we had quite a few problems to overcome, but for the last year things have been very stable. We also serve 40k users.
Mix sounds promising and I think the same setup I've outlined above could be done with UW *if* you separate mailboxes from home directories.
-- ☮ Sebastian Hagedorn ✉ [EMAIL PROTECTED] ❧ http://www.uni-koeln.de/~a0620/ _______________________________________________ Imap-uw mailing list [email protected] https://mailman1.u.washington.edu/mailman/listinfo/imap-uw
