[
https://issues.apache.org/jira/browse/HADOOP-13190?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15412255#comment-15412255
]
Xiao Chen commented on HADOOP-13190:
------------------------------------
Thank you [~jojochuang] for creating this jira and posting a patch!
We should definitely document this, and overall looks good. Comments:
- Maybe we can change the current {{$H3}} level title to {{Using Multiple
Instances of KMS}}, and list the current LB/VIP and the new LBKMSCP under it?
The other sub-sections (kerberos, secret-sharing) applies to multiple instances
in general.
- In the new LBKMSCP section, please also add the failure-handling behavior. If
a request to a KMSCP failed, LBKMSCP will retry the next KMSCP. The request is
returned as failure only if all KMSCPs failed.
- In the sample xml, maybe also list an http example?
> LoadBalancingKMSClientProvider should be documented
> ---------------------------------------------------
>
> Key: HADOOP-13190
> URL: https://issues.apache.org/jira/browse/HADOOP-13190
> Project: Hadoop Common
> Issue Type: Improvement
> Components: documentation, kms
> Affects Versions: 2.7.0
> Reporter: Wei-Chiu Chuang
> Assignee: Wei-Chiu Chuang
> Labels: supportability
> Attachments: HADOOP-13190.001.patch
>
>
> Currently, there are two ways to achieve KMS HA.
> The first one, and the only documented one, is running multiple KMS instances
> behind a load balancer.
> https://hadoop.apache.org/docs/stable/hadoop-kms/index.html
> The other way, is make use of LoadBalancingKMSClientProvider which is added
> in HADOOP-11620. However the usage is undocumented.
> I think we should update the KMS document to introduce
> LoadBalancingKMSClientProvider, provide examples, and also update
> kms-site.xml to explain it.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
