[
https://issues.apache.org/jira/browse/YARN-3021?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14308422#comment-14308422
]
Vinod Kumar Vavilapalli commented on YARN-3021:
-----------------------------------------------
Though the patch unblocks the jobs in the short term, it seems like long term
this is still bad. Applications that want to run for longer than 7 days in such
setups will just fail without any other way.
May be the solution is the following:
- Explicitly have an external renewer system that has the right permissions to
renew these tokens. Working with such an external renewer system needs support
in frameworks, for e.g. in MapReduce, a renewal server list similar to
mapreduce.job.hdfs-servers.
- RM can simply inspect the incoming renewer specified in the token and skip
renewing those tokens if the renewer doesn't match it's own address. This way,
we don't need an explicit API in the submission context.
Apologies for going back and forth on this one. Does that work? /cc [~jianhe],
[~kasha].
Irrespective of how we decide to skip tokens, the way the patch is skipping
renewal will not work. In secure mode, DelegationTokenRenewer drives the app
state machine. So if you skip adding the app itself to DTR, the app will be
completely stuck.
> YARN's delegation-token handling disallows certain trust setups to operate
> properly
> -----------------------------------------------------------------------------------
>
> Key: YARN-3021
> URL: https://issues.apache.org/jira/browse/YARN-3021
> Project: Hadoop YARN
> Issue Type: Bug
> Components: security
> Affects Versions: 2.3.0
> Reporter: Harsh J
> Attachments: YARN-3021.001.patch, YARN-3021.002.patch,
> YARN-3021.003.patch, YARN-3021.patch
>
>
> Consider this scenario of 3 realms: A, B and COMMON, where A trusts COMMON,
> and B trusts COMMON (one way trusts both), and both A and B run HDFS + YARN
> clusters.
> Now if one logs in with a COMMON credential, and runs a job on A's YARN that
> needs to access B's HDFS (such as a DistCp), the operation fails in the RM,
> as it attempts a renewDelegationToken(…) synchronously during application
> submission (to validate the managed token before it adds it to a scheduler
> for automatic renewal). The call obviously fails cause B realm will not trust
> A's credentials (here, the RM's principal is the renewer).
> In the 1.x JobTracker the same call is present, but it is done asynchronously
> and once the renewal attempt failed we simply ceased to schedule any further
> attempts of renewals, rather than fail the job immediately.
> We should change the logic such that we attempt the renewal but go easy on
> the failure and skip the scheduling alone, rather than bubble back an error
> to the client, failing the app submission. This way the old behaviour is
> retained.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
