Andrey created ZOOKEEPER-2400:
---------------------------------
Summary: ZooKeeper not starting: Follower is ahead of the leader
Key: ZOOKEEPER-2400
URL: https://issues.apache.org/jira/browse/ZOOKEEPER-2400
Project: ZooKeeper
Issue Type: Bug
Components: quorum
Affects Versions: 3.4.6
Reporter: Andrey
Steps to reproduce:
# Select deprecated algorithm in zoo.cfg: {code}electionAlg=0{code}
# Start zookeeper cluster: A(index 3),B(index 1),C(index 2) nodes
# Stop A node.
# Make some change to zk data. i.e. re-create ephemeral node. Make sure
currentEpoch increased in B and C nodes.
# currentEpoch/acceptedEpoch in node A less than B/C epoch
# Stop node B. Zookeeper cluster is not available
# Start node A. In A's node logs:
{code}
LEADING [quorum.QuorumPeer] [QuorumPeer[myid=3]/0:0:0:0:0:0:0:0:15523]
LEADING - LEADER ELECTION TOOK - 1458721180995 [quorum.Leader]
Follower sid: 2 : info :
org.apache.zookeeper.server.quorum.QuorumPeer$QuorumServer@3a888c1
java.io.IOException: Follower is ahead of the leader, leader summary: 10
(current epoch), 42949672964 (last zxid)
at
org.apache.zookeeper.server.quorum.Leader.waitForEpochAck(Leader.java:894)
at
org.apache.zookeeper.server.quorum.LearnerHandler.run(LearnerHandler.java:365)
...
Follower sid: 1 : info :
org.apache.zookeeper.server.quorum.QuorumPeer$QuorumServer@5001b9f5
...
java.lang.InterruptedException: Timeout while waiting for epoch to be acked by
quorum
at
org.apache.zookeeper.server.quorum.Leader.waitForEpochAck(Leader.java:915)
at org.apache.zookeeper.server.quorum.Leader.lead(Leader.java:394)
at
org.apache.zookeeper.server.quorum.QuorumPeer.run(QuorumPeer.java:799)
{code}
The logs above will be printed indefinitely and cluster won't recover.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
