We have this usecase where we want to bring down our live cluster,say
cluster#1 (serving online traffic) for machine upgradation, and route the
traffic during the upgrade to a backup cluster, say cluster#2
However, we want to drain the replication queue on cluster#1 before shutting
it down completely. In other words we want the replication to be over on
cluster#1 before we start the upgrade process.
I'm thinking about using the ReplicationZookeeper APIs to poll/watch the
appropriate the RS znodes to do this(Haven't thought about the complete
implementation details yet, just figuring out the feasibility part).
What do you guys think about it? Are there any caveats I should be aware of?
View this message in context: http://apache-hbase.679495.n3.nabble.com/Replication-Tracking-the-replication-status-of-a-cluster-tp4033878.html
Sent from the HBase User mailing list archive at Nabble.com.