Remote mailbox creation error / unified murder

[Janne Peltonen]

On Sun, Jul 01, 2007 at 02:01:11AM +0300, Janne Peltonen wrote:
> ..the prefork numbers. Now this is something. I don't really know which one of
> these is the one that matters, what is the resource that gets exhausted, why it
> gets exhausted etc. (and I sure as <expletive deleted> hope the problem only
> manifests with a large prefork number, not when my system is in production and
> has many more than 25 simultaneous imapds... well, it's easy to test...)

OK. I should only run 1 mupdate master process at a time. I'm sure this
is documented somewhere.

But when try to imapsync the old server's data to the new one, with only
one multi-threading mupdate master process, I keep getting 'connection
to the master timed out' errors from the mupdate slaves. I tried
increasing the relevant variables:

mupdate_connections_max: 1024
mupdate_workers_max: 500
mupdate_workers_maxspare: 100
mupdate_workers_minspare: 20
mupdate_workers_start: 50

but the errors only disappeared after I restarted the mupdate master
with the 35 preforked mupdate master processes. Now I wonder what would
be good values for those options when the system is in production, when
I have to run only one mupdate master process (or else create trouble
with mailbox list manipulation and proxying).  There'll probably be much
less mupdate load, since the amount of mailbox manipulation will be much
less...

I'll have some 50000 (fifty thousand) users with some 600000 (six
hundred thousand) mailboxes, with max abt 6000 users active at any given
time. And the murder has 24 nodes.


--Janne
-- 
Janne Peltonen <janne.peltonen at helsinki.fi>