Apache ZooKeeper sometimes fails to create a data directory on startup

Description

The Grid Service Agent failed to start the ZooKeeper instance, but succeeded on its next attempt. The failure was caused by a file system error trying to create the directory structure for the snapshot data and logs of the ZooKeeper server.

For example, the following message appears in the manager logs:
"Failed to run Zookeeper instance; Caused by: java.io.IOException: Unable to create data directory /work/manager/zookeeper/log/version-2"

This does not always occur, and there have been reports of similar behavior by the ZooKeeper community. To avoid this issue, the XAP Manager now creates these directories beforehand with write access, or lets the Grid Service Agent succeed in creating the ZooKeeper server on its next attempt.

Workaround

Create the directory manually with write access or let the Grid Service Agent succeed on its next attempt to create the ZooKeeper instance.

Acceptance Test

manager suite running CLI tests

Status

Assignee

Meron Avigdor

Reporter

Meron Avigdor

Labels

None

Priority

Medium

SalesForce Case ID

None

Fix versions

Commitment Version/s

None

Due date

None

Product

XAP

Edition

Premium

Platform

All

Sprint

None
Configure