It turns out that a ton of lists are exhibiting the too-large-date index. All date indexes are now being rebuilt; the process will take a few hours. In the meantime all incoming mails are being added to the (already rather enormous) queue.
The service has upgraded to mhonarc 2.5.0 again. Let's see what happens. On tap for later today: * deletion of YahooGroups archives (as discussed) * manual examination of listnames looking for misfilings * adjustment of FAQ to reflect EXIM's reject filter * adjustment of secondary MX to reflect primary MX policy * watching to see what, if anything, still gums up the works Assuming I can make headway today on performance, then I think this is also the time to think about another hardware iteration (in another n months). RAID5 is too slow, SCSI is too expensive, SMP is too finicky. Too bad RAM maxes out at ~1.5 or 2GB. HWRAID5/SCSI/ext2 -> lvm/IDE/reiserfs Debian2.2/Linux2.2 -> Debian3.0/Linux2.4 "mature" 2-CPU -> "bleeding edge" 1-CPU 186GB/1GB -> 1TB/1.5GB It may also be time to think about quality of service issues and if I ever want to break mail-archive into a regular/premium variants. -Jeff _______________________________________________ Gossip mailing list [EMAIL PROTECTED] http://jab.org/cgi-bin/mailman/listinfo/gossip
