zookeeper-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Vladimir Ivić (JIRA) <j...@apache.org>
Subject [jira] [Created] (ZOOKEEPER-3398) Learner.connectToLeader() may take too long to time-out
Date Tue, 21 May 2019 01:40:00 GMT
Vladimir Ivić created ZOOKEEPER-3398:
----------------------------------------

             Summary: Learner.connectToLeader() may take too long to time-out 
                 Key: ZOOKEEPER-3398
                 URL: https://issues.apache.org/jira/browse/ZOOKEEPER-3398
             Project: ZooKeeper
          Issue Type: Improvement
          Components: leaderElection, quorum
            Reporter: Vladimir Ivić
             Fix For: 3.6.0


After leader election happens, the followers will connect to the leader which is facilitated
by the Learner.connectToLeader() method. 

Learner.connectToLeader() is relying on the initLimit configuration value to time-out in case
the network connection is unreliable. This config may have a high value that could leave the
ensemble retrying and waiting in the state of not having quorum for too long. The follower
will retry up to 5 times. 

This patch introduces a new configuration directive that will allow Zookeeper to use different
time-out value `connectToLeaderLimit` which then could be set to lower value than `initLimit`.

 



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

Mime
View raw message