Artemis - cluster reconnection

["anton.mithun" <mi...@gmail.com>]

Hi,
I have implemented a cluster of two servers using udp discovery. 
udp://231.7.7.7:9876?ha=true&retryInterval=1000&retryIntervalMultiplier=1.0&reconnectAttempts=-1 
with one master and the other slave.

and also ran application with only one broker (no cluster)
tcp://localhost:61616?ha=true&retryInterval=1000&retryIntervalMultiplier=1.0&reconnectAttempts=-1

reconnection/failover only happens when having only one server. It doesn't
happen in a master-slave structure(former case).
When slave goes down, the slave will take over. And when the master is up
again, the failover will occur. But when both master and slave are down, it
tries to reconnect i suppose. But reconnection never happens even if two
servers are back.
is it like there is no reconnection strategy in master-slave cluster
enviroment?do we just have to restart the client after restarting the two
servers? or is it something wrong with my broker conf?



--
View this message in context: http://activemq.2283324.n4.nabble.com/Artemis-cluster-reconnection-tp4715113.html
Sent from the ActiveMQ - Dev mailing list archive at Nabble.com.

Re: Artemis - cluster reconnection

[Justin Bertram <jb...@apache.com>]

I didn't really understand your original use-case, but what you're saying about replication masters and slaves is true.  A replicated slave will not start without being able to contact its master since it doesn't know if it has up-to-date data.  This is noted in the documentation [1].


Justin

[1] http://activemq.apache.org/artemis/docs/1.3.0/ha.html (see the "Data Replication" section)

----- Original Message -----
From: "anton.mithun" <mi...@gmail.com>
To: dev@activemq.apache.org
Sent: Friday, August 12, 2016 12:33:05 AM
Subject: Re: Artemis - cluster reconnection

Okay found out something during my experiments. 
When in master slave - replication mode, then we have to know which has the
latest data. 
If both master and slave are down. Then configure the one with the latest
data as the master and the other one slave.
Then turn them back on and the re-connection occurs properly. 

Other wise we have to deploy shared storage strategy, where this is all done
automatically i guess.



--
View this message in context: http://activemq.2283324.n4.nabble.com/Artemis-cluster-reconnection-tp4715113p4715439.html
Sent from the ActiveMQ - Dev mailing list archive at Nabble.com.

Re: Artemis - cluster reconnection

["anton.mithun" <mi...@gmail.com>]

Okay found out something during my experiments. 
When in master slave - replication mode, then we have to know which has the
latest data. 
If both master and slave are down. Then configure the one with the latest
data as the master and the other one slave.
Then turn them back on and the re-connection occurs properly. 

Other wise we have to deploy shared storage strategy, where this is all done
automatically i guess.



--
View this message in context: http://activemq.2283324.n4.nabble.com/Artemis-cluster-reconnection-tp4715113p4715439.html
Sent from the ActiveMQ - Dev mailing list archive at Nabble.com.