[ACCUMULO-2224] ZooSession should be more robust to transient DNS issues

• Type: Bug
• Status: Resolved
• Priority: Minor
• Resolution: Fixed
• Affects Version/s: 1.4.1, 1.4.2, 1.4.3, 1.4.4, 1.5.0
• Fix Version/s: 1.4.5, 1.5.1, 1.6.0
• Component/s: client

Environment

1.4.5-SNAP on CDH4 w/gremlins

• Assignee: Sean Busbey
• Reporter: Sean Busbey
• Created: 20/Jan/14 17:30
• Updated: 01/Jul/14 15:46
• Resolved: 23/Jan/14 13:19

Description

While injecting network faults, I found that transient DNS problems caused us to bail out of ZooSessions rather than retrying as we do for all other IO problems. We should retry these failures just as we do for Connection Refused or other networking problems.

Since the addition of ACCUMULO-131, we can be sure that we won't retry actual invalid hosts for ever. Instead, after the time out period that holds for all other problems we'll properly exit.

The warn messages logged for IOExceptions should suffice to indicate improperly specified host names.

Issue Links

• breaks: ACCUMULO-2967 Unknown Host results in unending wait for clients
• links to: review board