[ACCUMULO-1923] Ensure ZK instances closed correctly

• Type: Bug
• Status: Resolved
• Priority: Minor
• Resolution: Won't Fix
• Component/s: client

Environment

development single node, hadoop 2.2,

• Assignee: Bill Havanki
• Reporter: Eric C. Newton
• Created: 22/Nov/13 22:00
• Updated: 03/Jan/14 01:18
• Resolved: 03/Jan/14 01:18

Description

Was: "warning at the end of a MapReduce job"

Ensure that ZK instances are closed where opened across the Accumulo codebase. This is an expansion of the original ticket scope, whose description is below.

Original Description:

While running the randomwalk tests I saw this at the end of a map/reduce job:

22 16:56:30,583 [client.ZooKeeperInstance] WARN : ZooKeeperInstance being cleaned up without being closed. Please remember to call close() before dereferencing to clean up threads.

This came just after the warning:

22 16:56:31,742 [sequential.MapRedVerify] DEBUG: Dropping table: sequential_hostname_1385157013877_MR

Issue Links

• breaks: ACCUMULO-2105 [RW] Multitable.CopyTable failed from using a closed ZooKeeperInstance
• is superceded by: ACCUMULO-2128 Provide resource cleanup via static utility rather than Instance.close