Minor Description
Here's what I get:
Testcase: testLeaderInConnectingFollowers took 34.117 sec
Testcase: testLastAcceptedEpoch took 0.047 sec <----- new test added in ZK-1343
Testcase: testLeaderInElectingFollowers took 0.004 sec
Caused an ERROR
Address already in use
java.net.BindException: Address already in use
at java.net.PlainSocketImpl.socketBind(Native Method)
at java.net.PlainSocketImpl.bind(PlainSocketImpl.java:383)
at java.net.ServerSocket.bind(ServerSocket.java:328)
at java.net.ServerSocket.<init>(ServerSocket.java:194)
at java.net.ServerSocket.<init>(ServerSocket.java:106)
at org.apache.zookeeper.server.quorum.Leader.<init>(Leader.java:220)
at org.apache.zookeeper.server.quorum.Zab1_0Test.createLeader(Zab1_0Test.java:711)
at org.apache.zookeeper.server.quorum.Zab1_0Test.testLeaderInElectingFollowers(Zab1_0Test.java:225)
Testcase: testNormalFollowerRun took 29.128 sec
Testcase: testNormalRun took 25.158 sec
Testcase: testLeaderBehind took 25.148 sec
Testcase: testAbandonBeforeACKEpoch took 34.029 sec
My guess is that testLastAcceptedEpoch doesn't properly close the connection before testLeaderInElectingFollowers starts.
I propose to add
if (leadThread != null) {
leadThread.interrupt();
leadThread.join();
}
to the test.
In addition, I propose to change the hard-wired ports in Zab1_0Test to use Portassignment.unique() as done in other tests. If I understand correctly the static counter used in unique() to assign ports is initialized once per test file, so it would also prevent the problem I'm seeing here of two tests in the same file trying to use the same port.
The error can be reproduced using the attached patch (for some reason I don't see the problem in the trunk).