[
https://issues.apache.org/jira/browse/HBASE-19524?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16294633#comment-16294633
]
Duo Zhang commented on HBASE-19524:
-----------------------------------
For TestMasterReplication, it is a test case issue. For testCyclicReplication2,
the last assert
{code}
wait(row4, htables[0], true);
{code}
should be
{code}
wait(row4, htables[0], false);
{code}
Not sure why it can pass on master, at least it should be flakey... The last
parameter means whether this is a delete, so if it is true, then we will check
if the row is disappeared, which could also pass if we have not replicated the
row yet...
All the replication related operations are asynchronous so the UT is not very
stable and may not test what we want. After HBASE-19397 I think we can have
better result as we can wait on the procedure to finish.
Let me open a new issue for TestMasterReplication.
> Master side changes for moving peer modification from zk watcher to procedure
> -----------------------------------------------------------------------------
>
> Key: HBASE-19524
> URL: https://issues.apache.org/jira/browse/HBASE-19524
> Project: HBase
> Issue Type: Sub-task
> Components: proc-v2, Replication
> Reporter: Duo Zhang
> Assignee: Duo Zhang
> Fix For: HBASE-19397
>
> Attachments: HBASE-19524-HBASE-19397-v1.patch,
> HBASE-19524-HBASE-19397-v2.patch, HBASE-19524-HBASE-19397.patch
>
>
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
