Hi,
It seems the author is no more working on it, so I wouldn't invest too
much time in this option. As you said, it was designed on James 2.x,
which works with database for spooling, meaning there's a lot of
concurrency concerns/issues to solve.
With James 3.x, you've got a more distributable enabled platform:
- The spool: we can have a distributed network of brokers based on active-mq
- The mailbox store: we are now working on HBase for
mailbox/users/domains, other distributed mailboxes exist and may be one
day released opensource
- The Uid generation (needed for IMAP): we need to have a distributed
generation/cache implementation for this
Once this is done, we need to explain how to configure it, and also
provide tools to manage/monitor.
So Jame3-HA is not yet here today, and we don't have timeline for it,
although we are pretty confident that it will come one day...
Hope this helps.
- Eric
On 21/06/11 06:02, SuoNayi wrote:
I can not contact he via email,my email to his address is fallen back and it
seems his email domain does not exist anymore.
I am banging my head on the door too now~
Help~
At 2011-06-21??"Norman Maurer"<[email protected]> wrote:
I guess you would better ask the author of it. I dont know anyone who use it.
bye
norman
Am Montag, 20. Juni 2011 schrieb SuoNayi<[email protected]>:
Ths Norman,I have known that james 3.x supports distribution because of using
AMQ queue to be its spool repository.
What I care about is that does james-ha for james 2.x work?
At 2011-06-20??"Norman Maurer"<[email protected]> wrote:
Hi there,
I never used it. But I'm currently working on getting plain James 3.0
ready for clustering. I think its not far away from complete.
Bye,
Norman
2011/6/19 SuoNayi<[email protected]>:
Hi all, it's known that james-ha provides a high availability distribution of
James via the distribution lock.
But I cannot understand that:
for example, I have two james cluster member and there are 20 mail in the
SpoolRepository to be dispatched.When two member start at the same
time,JamesSpoolManager begins to read mails from the same database. each
cluster member obtains its own spool mails. these pending mails are the same
because of the same database, 20 mail for example.each cluster member has 20
mails in its own internal pending queue.two cluster member start to dispatch
mails.member 1 is faster than member 2.so member 1 is processing the 10th mail
in its own spool, while member 2 is processing the 2th mail in its spool.member
2 may redispath mails sent by member 1 because member 2 can obtain locks
released by member 1 and the locks released by member 1 has been processed.
In one word,because cluster member has its own spool which is not synchronized
with others,they can obtain the same lock.
That's all,I hope someone can make me clearly,Ths a lot!
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
--
Eric
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
