[
https://issues.apache.org/jira/browse/YARN-1029?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13861990#comment-13861990
]
Vinod Kumar Vavilapalli commented on YARN-1029:
-----------------------------------------------
Some comments/questions on the last patch:
- yarn_server_resourcemanager_service_protos.proto: RMActiveNodeInfoProto ->
ActiveRMInfoProto ?
- yarn-default.xml: "This kind of failover is embedded in the RM and does not
explicitly fence stores.” - “does not” or “does”?
- I think we should force admins to set yarn.resourcemanager.cluster-id
explicitly (only in case HA is enabled for now). Defaults don’t tend to be
changed and a default cluster-id can potentially cause hard-to-debug issues.
- No need for YarnBadConfigurationException. It isn’t adding any value and is
inconsistent with how we tackle misconfigs everywhere. Let’s just use
YarnRuntimeException.
- Why is ZK added to hadoop-yarn-client module? It should be only in
server-common?
- RMFatalEventType.EMBEDDED_ELECTOR -> EMBEDDED_ELECTOR_FAILED or something
like that? Similarly STORE_FENCED to STATE_STORE_FENCED and STORE_OP_FAILED to
STATE_STORE_OP_FAILED for making it explicit.
EmbeddedElectorService
- Initialized in AdminService? It can be initialize in ResourceManager class
itself and it can access AdminService via RMContext.
- It can similarly access rmDispatcher from RMContext.
Testing
- We should have one test that switches off the automatic failover. May be
retain the old testExplicitFailover test in TestRMFailover?
- TestRMHA.testTransitionsWhenAutomaticFailoverEnabled: After each transition,
check the state?
> Allow embedding leader election into the RM
> -------------------------------------------
>
> Key: YARN-1029
> URL: https://issues.apache.org/jira/browse/YARN-1029
> Project: Hadoop YARN
> Issue Type: Sub-task
> Reporter: Bikas Saha
> Assignee: Karthik Kambatla
> Attachments: embedded-zkfc-approach.patch, yarn-1029-0.patch,
> yarn-1029-0.patch, yarn-1029-1.patch, yarn-1029-2.patch, yarn-1029-3.patch,
> yarn-1029-4.patch, yarn-1029-5.patch, yarn-1029-6.patch, yarn-1029-7.patch,
> yarn-1029-7.patch, yarn-1029-8.patch, yarn-1029-9.patch,
> yarn-1029-approach.patch
>
>
> It should be possible to embed common ActiveStandyElector into the RM such
> that ZooKeeper based leader election and notification is in-built. In
> conjunction with a ZK state store, this configuration will be a simple
> deployment option.
--
This message was sent by Atlassian JIRA
(v6.1.5#6160)
