[ 
https://issues.apache.org/jira/browse/ZOOKEEPER-3818?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17100660#comment-17100660
 ] 

Mate Szalay-Beko commented on ZOOKEEPER-3818:
---------------------------------------------

well, maybe the release notes should be more precise :)
all 3.5.x after 3.5.5 is considered as production stable. There are relatively 
few changes in the bugfix releases, but still each bugfix release contains 
important security and other fixes. So I would recommend to use the 
3.5.<latest> in production.

About 3.6 I am not sure. I know that some in the community are using it in 
production, but I think they are still in minority.

> zkServer.sh status command exits if clientPort is missing even if 
> secureClientPort is present for SSL
> -----------------------------------------------------------------------------------------------------
>
>                 Key: ZOOKEEPER-3818
>                 URL: https://issues.apache.org/jira/browse/ZOOKEEPER-3818
>             Project: ZooKeeper
>          Issue Type: Bug
>          Components: security
>    Affects Versions: 3.5.5
>            Reporter: Aishwarya Soni
>            Assignee: Mate Szalay-Beko
>            Priority: Critical
>
> I am configuring SSL on Zookeeper 3.5.5 branch and have removed the 
> clientPort config from zoo.cfg and adding onlysecureClientPort. Also, I have 
> removed it from my server ensemble connection string in zoo.cfg.dynamic file 
> as it results in a port binding issue on the port 2181 if we keep it in both 
> the files.
> But, in zkServer.sh, it checks if the clientPort is set in the *status* cmd 
> else it throws exit 1 and terminates the process. How to overcome this 
> situation? We cannot see the clientPort in zoo.cfg as it would enable mixed 
> mode which we do not want when we enable SSL.
> Also, I am using zkServer.sh status output as a healthcheck for our 
> containerized zookeeper to see if thee quorum is established or not as in 
> cluster mode, zookeeper can finally run either in follower or leader state 
> (ignoring intermediate state changes). So as the status output throws exit 1, 
> the healthcheck is also failing.



--
This message was sent by Atlassian Jira
(v8.3.4#803005)

Reply via email to