Nick Ustinov schrieb: > > We are planning to switch production server from 2.0.16 to 2.1.3/skiplist. > The system has about 600000 mailboxes and is quite heavily loaded (seeing > alot of db lockers). Are there any hidden catches people i should avoid? > Besides converting mailboxes file using cvt_cyrusdb is there anything else I > should know about? How about sieve scripts? Are these compatbile? Also, is > it recommended to switch to skiplist only for mailbox file or for the rest > as well? > > Sincerely, > Nick
Hi Nick, this is a cut and paste from a mail where I asked about skiplist: > >>>>> On Thu, 21 Mar 2002 07:40:30 +0100, > >>>>> Simon Matter <[EMAIL PROTECTED]> (sm) writes: > > sm> OT, but did you upgrade all dbs from DB3 to skiplist? Is it possible at > sm> all? I'm not sure about this but as I understand the configure script it > sm> should be possible now to use skiplist for every db. So my question is > sm> whether this is true, if yes then it means we can compile cyrus without > sm> depending on any DB[3,4] whathever, right? > sm> I'd like this idea because building RPM (or other bin packages) is much > sm> easier if you don't depend on too many libs. > > At this time I believe it would not be good to replace deliver.db or > tls.db with skiplist. While skiplist works well for mailboxes.db, I > don't think that would be the case for deliver.db or tls.db. There > are many more writes going on to these than mailboxes.db. HTH -Simon
