hadoop-yarn-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Zhijie Shen (JIRA)" <j...@apache.org>
Subject [jira] [Updated] (YARN-1776) renewDelegationToken should survive RM failover
Date Thu, 20 Mar 2014 04:58:43 GMT

     [ https://issues.apache.org/jira/browse/YARN-1776?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]

Zhijie Shen updated YARN-1776:
------------------------------

    Attachment: YARN-1776.1.patch

I created a patch:

1. Add updateRMDelegationTokenAndSequenceNumber to RMStateStore.

2. For MemoryRMStateStore, we don't need to make the method atomic as the memory is lost when
RM fails. Therefore, it just a simple wrapper of storeRMDelegationTokenAndSequenceNumber and
removeRMDelegationToken.

3. For ZKRMStateStore,  I make use of opList to group both delete and store operations together,
to ensure all or no operations get succeeded.

4. For FileSystemRMStateStore, it is a difficult case: since we're not just touching a single
file, it's hard to make all or no fs operations succeed. Therefore, I just leave it as what
I've done for MemoryRMStateStore. Meanwhile, storeRMDelegationTokenAndSequenceNumber itself
is not atomic as well.

The good thing, is that RM failover is supposed to work with ZK impl. Hopefully it is still
OK. Thoughts?

5. RMDelegationTokenSecretManager#updateStoredToken calls updateRMDelegationTokenAndSequenceNumber
then.

6. Add the test for updateRMDelegationTokenAndSequenceNumber.

> renewDelegationToken should survive RM failover
> -----------------------------------------------
>
>                 Key: YARN-1776
>                 URL: https://issues.apache.org/jira/browse/YARN-1776
>             Project: Hadoop YARN
>          Issue Type: Sub-task
>            Reporter: Zhijie Shen
>            Assignee: Zhijie Shen
>         Attachments: YARN-1776.1.patch
>
>
> When a delegation token is renewed, two RMStateStore operations: 1) removing the old
DT, and 2) storing the new DT will happen. If RM fails in between. There would be problem.



--
This message was sent by Atlassian JIRA
(v6.2#6252)

Mime
View raw message