hbase-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "stack (JIRA)" <j...@apache.org>
Subject [jira] [Created] (HBASE-19261) ClusterStatusPublisher where Master could optionally broadcast notice of dead servers is broke
Date Wed, 15 Nov 2017 06:13:00 GMT
stack created HBASE-19261:
-----------------------------

             Summary: ClusterStatusPublisher where Master could optionally broadcast notice
of dead servers is broke
                 Key: HBASE-19261
                 URL: https://issues.apache.org/jira/browse/HBASE-19261
             Project: HBase
          Issue Type: Bug
          Components: MTTR
            Reporter: stack


An old feature that you could optionally enable so Master would broadcast via UDP/Multicast
messages to registered clients when it learned of dead servers is broken in hbase2. It was
added here:

commit c352a848a1af318df4f0348728fb136ce289d3e4
Author: nkeywal <nkeywal@unknown>
Date:   Tue Mar 19 10:07:17 2013 +0000

    HBASE-7590 Add a costless notifications mechanism from master to regionservers & clients
- new files

It could help the recalibration of clients on server death run quicker. I have not stats on
how often it was use if at all.

I tried playing around with it to try and get it going again but need to spend more time on
it. Putting aside for now. 

We should make it work again and then talk it up as a mechanism that can help improve MTTR
(or strip it).



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)

Mime
View raw message