On Fri, January 6, 2012 11:04 am, Sebastian Hagedorn wrote: > Hi Pascal, > > > thanks for your reply! I decided to ask my more specific questions in separate > mails, but I have a few follow-up questions for you. > > --On 5. Januar 2012 19:50:28 +0100 Pascal Gienger > <pascal.gienger@xxxxxxxxxxxxxxx> wrote: > > >> In principle it is simple, but I hit a hard ground today. I did the >> mistake to miscalculate the "converting indexes".... >> >> Our storage was perfectly sized to handle the needed IOPS for cyrus >> operation, but having 8000 index files being rebuild simultaneosuly... > > I am very aware of that issue, but we haven't yet decided how to handle it. Sebastian, Pascal, Neither have we, and I know for sure my I/O subsystem is unable to handle the extra metadata reconstruction load. To give you an idea : 73k users, 508k mailboxes, 4.5 TB of messages, 48M messages. In an additional effort to grasp the dynamics of our central mail system (when hosting over 10k users it pays to have regular reporting and statistics produced) I discovered that no less than 66% of messages reside in Inboxes. Which is extraordinary, given the fact that POP3 users number less than 10% of the total. (We are a mixed IMAP/POP shop, for historical reasons.) If more messages were to be found in sub-mailboxes, metadata conversion load would have been more distributed in time, as mailboxes are accessed. Regards, Eric Luyten, Computing Centre VUB/ULB. ---- Cyrus Home Page: http://www.cyrusimap.org/ List Archives/Info: http://lists.andrew.cmu.edu/pipermail/info-cyrus/
