You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@hbase.apache.org by "ruanhui (Jira)" <ji...@apache.org> on 2022/09/08 03:18:00 UTC

[jira] [Commented] (HBASE-27304) Support using IP to expose master/rs servers for some special scenarios

    [ https://issues.apache.org/jira/browse/HBASE-27304?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17601598#comment-17601598 ] 

ruanhui commented on HBASE-27304:
---------------------------------

Nice feature!

We encountered the same problem when deploying HBase with docker. Two docker containers can not resolve each other's hostname, and we can not reset hostname to IP for safety reasons. After reading the code, we used a somewhat tricky approach to solve this problem: set hbase.unsafe.regionserver.hostname and hbase.master.hostname to a special value and use sed command to replace this special value with the host ip in dockerfile, the command is like this :

sed -i "s/hbase.unsafe.regionserver.hostname.value/$(hostname --all-ip-addresses | sed 's/[ \t]*$//g')/g" hbase-site.xml

We are also planning to develop a new feature to solve this problem. It's really nice to see this issue. (y)

> Support using IP to expose master/rs servers for some special scenarios
> -----------------------------------------------------------------------
>
>                 Key: HBASE-27304
>                 URL: https://issues.apache.org/jira/browse/HBASE-27304
>             Project: HBase
>          Issue Type: New Feature
>          Components: master, regionserver
>    Affects Versions: 3.0.0-alpha-4
>            Reporter: LiangJun He
>            Assignee: LiangJun He
>            Priority: Minor
>             Fix For: 3.0.0-alpha-4
>
>
> In our environment, when accessing the HBase cluster from another cluster (such as the Flink cluster), you cannot use the hostname to access the HBase cluster, but you can use IP. Because it is possible that the hostname of the HBase cluster automatic expansion node is not maintained in the Flink cluster. So we should support using IP to expose master/rs servers for this special scenarios.
>  



--
This message was sent by Atlassian Jira
(v8.20.10#820010)