[KNOX-546] WebHDFS HA secure cluster requests hang on revisiting a namenode during failover

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 0.6.0
• Fix Version/s: 0.7.0
• Component/s: Server
• Assignee: Sumit Gupta
• Reporter: Sumit Gupta
• Created: 21/May/15 20:04
• Updated: 28/Mar/19 14:15
• Resolved: 22/May/15 23:51

Description

When making a requests through knox to WebHDFS HA the following scenario does not work.

WebHDFS HA is setup with two nodes, one active and one in standby. WebHDFS HA is configured in knox with these two urls.

The active namenode is restarted (comes back up as standby node) and a request is made. A failover happens and the request is successful.
The new active node is then restarted so that the old active node is now active again. When another request is made the request hangs.

Please note that this scenario is only true in a secure cluster setup.