[
https://issues.apache.org/jira/browse/ASTERIXDB-1251?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15118272#comment-15118272
]
ASF subversion and git services commented on ASTERIXDB-1251:
------------------------------------------------------------
Commit 8fc8bf8b510bdc635f949f2eebf8b4d0d2a6b008 in incubator-asterixdb's branch
refs/heads/master from [~mhubail]
[ https://git-wip-us.apache.org/repos/asf?p=incubator-asterixdb.git;h=8fc8bf8 ]
Asterix NCs Fault Tolerance
This change includes the following:
- Adapt replication to unique partitions storage.
- Implement auto failover for failing NCs.
- Implement auto failover for metadata node.
- Fix for ASTERIXDB-1251 using proper error message.
- Basic replication test cases using vagrant virtual cluster for:
1. LSM bulkload components replication.
2. LSM Memory components replication and recovery.
3. Metadata node takeover.
These test cases will be part of the cluster test profile.
Change-Id: Ice26d980912a315fcb3efdd571d6ce88717cfea4
Reviewed-on: https://asterix-gerrit.ics.uci.edu/573
Tested-by: Jenkins <[email protected]>
Reviewed-by: Till Westmann <[email protected]>
Reviewed-by: abdullah alamoudi <[email protected]>
> NPE at NC startup after upgrade
> -------------------------------
>
> Key: ASTERIXDB-1251
> URL: https://issues.apache.org/jira/browse/ASTERIXDB-1251
> Project: Apache AsterixDB
> Issue Type: Bug
> Components: Metadata
> Reporter: Yingyi Bu
> Assignee: Murtadha Hubail
>
> I upgraded my instance to latest master but got this exception:
> {noformat}
> java.lang.NullPointerException
> at
> org.apache.asterix.metadata.bootstrap.MetadataBootstrap.enlistMetadataDataset(MetadataBootstrap.java:412)
> at
> org.apache.asterix.metadata.bootstrap.MetadataBootstrap.startUniverse(MetadataBootstrap.java:199)
> at
> org.apache.asterix.hyracks.bootstrap.NCApplicationEntryPoint.notifyStartupComplete(NCApplicationEntryPoint.java:235)
> at
> org.apache.hyracks.control.nc.NodeControllerService.start(NodeControllerService.java:311)
> at org.apache.hyracks.control.nc.NCDriver.main(NCDriver.java:48)
> {noformat}
> It seems related to the recent global resource id changes.
> But failing the instance forever seems not the right answer for that case. To
> me, the bottomline is that the instance can at least start (with warnings in
> the log saying which datasets are lost) though the old data could lose, so
> that a user doesn't need to reinstall a new instance.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
