[
https://issues.apache.org/jira/browse/HADOOP-14521?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16047207#comment-16047207
]
Xiao Chen commented on HADOOP-14521:
------------------------------------
Thanks for revving Rushabh. Looks pretty close to me.
Thanks for the explanations about {{AuthenticatedException}}, makes sense. I
don't know about why the IOExceptions are wrapped either (except for guessing
they're tricks with the method signature). Agree it's not related to this jira.
Patch 6 comments below:
For the retries, do you think maybe it's more intuitive to specify num of
failovers for the whole provider set? (i.e. {{maxFailovers =
conf.getInt(KMS_CLIENT_FAILOVER_MAX_RETRIES_KEY) * providers.length}}. (Then we
can have the core-default.xml set to 1) Will leave the decision to you and
Daryn.
Nits:
- KMSCP: the first {{@Link}} seems redundant in {{This will always create a
@Link ...}}
- Good to have some validations on the config params (e.g. numRetries > 0 etc.)
> KMS client needs retry logic
> ----------------------------
>
> Key: HADOOP-14521
> URL: https://issues.apache.org/jira/browse/HADOOP-14521
> Project: Hadoop Common
> Issue Type: Improvement
> Affects Versions: 2.6.0
> Reporter: Rushabh S Shah
> Assignee: Rushabh S Shah
> Attachments: HDFS-11804-trunk-1.patch, HDFS-11804-trunk-2.patch,
> HDFS-11804-trunk-3.patch, HDFS-11804-trunk-4.patch, HDFS-11804-trunk-5.patch,
> HDFS-11804-trunk-6.patch, HDFS-11804-trunk.patch
>
>
> The kms client appears to have no retry logic – at all. It's completely
> decoupled from the ipc retry logic. This has major impacts if the KMS is
> unreachable for any reason, including but not limited to network connection
> issues, timeouts, the +restart during an upgrade+.
> This has some major ramifications:
> # Jobs may fail to submit, although oozie resubmit logic should mask it
> # Non-oozie launchers may experience higher rates if they do not already have
> retry logic.
> # Tasks reading EZ files will fail, probably be masked by framework reattempts
> # EZ file creation fails after creating a 0-length file – client receives
> EDEK in the create response, then fails when decrypting the EDEK
> # Bulk hadoop fs copies, and maybe distcp, will prematurely fail
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
