[ https://issues.apache.org/jira/browse/SAMZA-1583?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16361626#comment-16361626 ]
Shanthoosh Venkataraman commented on SAMZA-1583: ------------------------------------------------ [~boryas] One possible fix is to generate JobModel using available processors and not fail the leader during JobModel generation phase. > Reading ephemeral node data throws ZkNoNodeExistsException. > ----------------------------------------------------------- > > Key: SAMZA-1583 > URL: https://issues.apache.org/jira/browse/SAMZA-1583 > Project: Samza > Issue Type: Bug > Reporter: Shanthoosh Venkataraman > Assignee: Shanthoosh Venkataraman > Priority: Major > > Leader reads the processorId's of zookeeper ephemeral nodes through the > following two operations to generate JobModel: > A. ephemeralZkNodes = Get the list of ephemeral zookeeper nodes. > B. For each zkNode in ephemeralZkNodes: > fetch data of the zkNode. > The zkNodes returned by the operation A, may no longer be available when > fetch data remote call was made. This will cause zkClient to throw > ZkNoNodeExistsException. This exception is unhandled currently, there by > failing the leader. -- This message was sent by Atlassian JIRA (v7.6.3#76005)