On Thu, 2012-03-22 at 11:19 +0100, Egoitz Aurrekoetxea wrote
I have a mailbox machine with near 6000 users. This machine is having
some delay on the replication. I'm debugging the replication and have
seen twice some problems with two users in they're seen database which
I solved basically removing the seen databases of this two users and
running a reconstruct -r -f (later seen invocations have worked for
one unless of this users). This machine has been created by doing
manual replications of two servers that where running cyrusimap 2.3.14
and this new one runs 2.3.18 which has the delays (although in
greetings sais is 2.3.16, I assume perhaps this has forgotten to be
changed in some banner field at code). Is there any advise for
speeding replication?. I have sync_repeat_interval to one second
(default). Is it normal replication to have some delay on perhaps busy
servers?. We have joined two servers in one and have a
master-slave replication now.
There are train loads of bug fixes since 2.3.x, many of which relate to
replication and database consistency. Not to mention that later 2.4.x
versions have a significant performance benefit.
signature.asc
Description: This is a digitally signed message part
Cyrus Home Page: http://www.cyrusimap.org/
List Archives/Info: http://lists.andrew.cmu.edu/pipermail/info-cyrus/
