cyrus replication validation

[John Capo]

Quoting Dmitriy Kirhlarov (dkirhlarov at oilspace.com):
> On Thu, Apr 05, 2007 at 12:10:14PM -0400, Ilya Vishnyakov wrote:
> 
> > Hello Cyrus Gurus!
> > I was wondering if there is any specific way to check if the
> > replication was done properly? I set up cyrus replication between two
> > servers (documentation I used:
> > http://cyrusimap.web.cmu.edu/imapd/install-replication.html). However,
> > before switching our production servers we would like to make sure
> > that replication was done properly. We checked if the directories are
> 
> On both servers:
> find imap/ -type f | awk '!/(cache|index|header)/ {print}' | sort > server1.lst
> find imap/ -type f | awk '!/(cache|index|header)/ {print}' | sort > server2.lst
> 
> and
> diff -u server1.lst server2.lst
> 

Quick mailboxes.db check.

ctl_mboxlist -d | md5   on server1
ctl_mboxlist -d | md5   on server2

Both hashes should be identical.  Or diff the ctl_mboxlist -d
outputs.

You should check the subscriptions on the replica too.  I don't
know of a simple way for you to verify the subscriptions other than
software that fetches and compares each each users subscriptions.
Subscription replication is the only replication problem I am seeing
these days and I haven't had time to look into it.

Well, that's not completely true.  I have seen some cases where the
bits controlling the POP3 UIDL format will differ on the replicas.
If all mailboxes were created fairly recently, for some value of
recent, or you have no POP3 users, you should not have a problem.

I have mailboxes that were originally created with early 1.X and
lots of POP3 users.  The UIDL format has changed over the years and
we have yet another UIDL format that attempts to get around the
Outlook problem.  The jury is still out on that.  The UIDL format
difference are only a problem if mail is left on server.

John Capo