[
https://issues.apache.org/jira/browse/HIVE-22992?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17054922#comment-17054922
]
Antal Sinkovits commented on HIVE-22992:
----------------------------------------
Hey [~prasanth_j], thanks for checking. I've run into this issue while I was
testing llap cache replication locally.
"If llap daemon restarts with same host name and port but with different
uniqueID (worker identity) does it not cache the new llap daemon?"
I don't think this will be an issue because of another bug. When a node
de-registers itself from zk, it will throw away all instances from the cache
running on that host.
https://github.com/apache/hive/blob/master/llap-client/src/java/org/apache/hadoop/hive/registry/impl/ZkRegistryBase.java#L430
Shall I create a follow up jira, for that issue?
Or there can be only 1 service per host, but in that case the Map<String,
Set<InstanceType>> nodeToInstanceCache seems to be misleading.
> ZkRegistryBase caching mechanism only caches the first instance
> ---------------------------------------------------------------
>
> Key: HIVE-22992
> URL: https://issues.apache.org/jira/browse/HIVE-22992
> Project: Hive
> Issue Type: Bug
> Components: llap
> Affects Versions: 4.0.0
> Reporter: Antal Sinkovits
> Assignee: Antal Sinkovits
> Priority: Minor
> Attachments: HIVE-22992.01.patch, HIVE-22992.02.patch
>
>
> ZkRegistryBase caching mechanism only caches the first instance of the llap
> node running on the same host.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
