[
https://issues.apache.org/jira/browse/HBASE-25216?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Duo Zhang updated HBASE-25216:
------------------------------
Description:
The failure of TestSeparateClientZKCluster is because that, we start the zk
syncer before we initialize meta region, and after HBASE-25099, we will scan
zookeeper to get the meta znodes directly instead of checking the config, so we
will get an empty list since there is no meta location on zk yet, thus we will
sync nothing.
But changing the order can not solve everything, as after HBASE-25099, we can
change the meta replica count without restartinig master, so the zk syncer
should have the ability to know the change and start to sync the location for
the new replicas.
> The client zk syncer should deal with meta replica count change
> ---------------------------------------------------------------
>
> Key: HBASE-25216
> URL: https://issues.apache.org/jira/browse/HBASE-25216
> Project: HBase
> Issue Type: Bug
> Components: master, Zookeeper
> Reporter: Duo Zhang
> Assignee: Duo Zhang
> Priority: Major
>
> The failure of TestSeparateClientZKCluster is because that, we start the zk
> syncer before we initialize meta region, and after HBASE-25099, we will scan
> zookeeper to get the meta znodes directly instead of checking the config, so
> we will get an empty list since there is no meta location on zk yet, thus we
> will sync nothing.
> But changing the order can not solve everything, as after HBASE-25099, we can
> change the meta replica count without restartinig master, so the zk syncer
> should have the ability to know the change and start to sync the location for
> the new replicas.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
