[HADOOP-9555] HA functionality that uses ZooKeeper may experience inadvertent TCP RST and miss session expiration event due to bug in client connection management

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 2.4.0, 3.0.0-alpha1
• Fix Version/s: 2.5.0
• Component/s: ha
• Target Version/s: 2.5.0
• Hadoop Flags: Reviewed
• Assignee: Chris Nauroth
• Reporter: Chris Nauroth
• Created: 09/May/13 20:31
• Updated: 12/May/16 18:21
• Resolved: 10/Jun/14 18:10

Description

ZOOKEEPER-1702 tracks a client connection management bug. The bug can cause an unexpected TCP RST that ultimately prevents delivery of a session expiration event. The symptoms of the bug seem to show up more frequently on Windows than on other platforms (though it's not really a Windows-specific bug).

Issue Links

• is depended upon by: HADOOP-9991 Fix up Hadoop POMs, roll up JARs to latest versions
• is related to: HADOOP-9556 disable HA tests on Windows that fail due to ZooKeeper client connection management bug
• is related to: HADOOP-10668 TestZKFailoverControllerStress#testExpireBackAndForth occasionally fails
• is related to: HADOOP-10754 Reenable several HA ZooKeeper-related tests on Windows.
• relates to: HADOOP-8562 Enhancements to support Hadoop on Windows Server and Windows Azure environments
• requires: ZOOKEEPER-1702 ZooKeeper client may write operation packets before receiving successful response to connection request, can cause TCP RST