[
https://issues.apache.org/jira/browse/HBASE-11746?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14098423#comment-14098423
]
oren razon commented on HBASE-11746:
------------------------------------
I understand your perspective but I think that if I set HBase retry to X (which
is >1) so as a user whenever I approach HBase my intent is that and sudden blip
connectivity issue won't cause me to any problem at the application level.
However in the state I was dealing with I had to dive into HBase code to
understand that for that specific case the configuration is hard coded to "1"
and therefor I should implement such retry on the application level.
I believe current state is not a big issue cause as a user I can do retry in
the application level but I think it's a bit misleading and for my case caused
me to several hours of work to understand why it work with only 1 retry
> checkHBaseAvailable method override {hbase.client.retries.number}
> configuration to be 1
> ---------------------------------------------------------------------------------------
>
> Key: HBASE-11746
> URL: https://issues.apache.org/jira/browse/HBASE-11746
> Project: HBase
> Issue Type: Bug
> Components: Admin
> Affects Versions: 0.98.3, 0.94.22
> Environment: ubuntu machine
> Reporter: oren razon
> Priority: Minor
> Labels: patch
> Fix For: 1.0.0, 0.94.23
>
> Original Estimate: 24h
> Remaining Estimate: 24h
>
> Hi,
> Although my HBase configuration is set for 10 for
> "hbase.client.retries.number" parameter, the checkHBaseAvailable method in
> org.apache.hadoop.hbase.client.HBaseAdmin class override this setting and try
> to connect master with only 1 retry.
> This setting cause us to HBase failure when HBase is not available for a
> sudden moment while checking for availability. Can it use the configured
> "hbase.client.retries.number" setting instead?
--
This message was sent by Atlassian JIRA
(v6.2#6252)
