[jira] [Commented] (AMBARI-4016) Ambari Server HA

["Hari Sekhon (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/AMBARI-4016?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16504465#comment-16504465 ] 

Hari Sekhon commented on AMBARI-4016:
-------------------------------------

Any update on this?

The details in https://issues.apache.org/jira/browse/AMBARI-7896 for ZooKeeper election may be useful as this is a similar design to other core ecosystem components.

> Ambari Server HA
> ----------------
>
>                 Key: AMBARI-4016
>                 URL: https://issues.apache.org/jira/browse/AMBARI-4016
>             Project: Ambari
>          Issue Type: New Feature
>          Components: ambari-server
>            Reporter: Peter Shi
>            Priority: Major
>              Labels: HA
>             Fix For: 2.3.0
>
>
> Our team has a fork of ambari 1.4.1, we need ambari service to be HA. 
> But seems currently not supported.
> Whether we can implement ambari HA like this
> 1)	Deploy 2 ambari-server with it database in 2 machine, say primary and backup;
> 2)	Sync data from primary database to backup database(using postgresql streaming sync) when have data updated.
> 3)	Use heartbeat between the 2 ambari-server to discover server_lost event, and make backup server to be primary; and the new backup server stop communicate with ambari-agent
> 4)	Each agent have 2 ambari-server’s address configurations(or negotiate from zookeeper?), once primary server changed, agent should register to new one.
> 5)	Backup webui can redirect page to primary one. GET method REST api can access both ambari-server
> My question
> a.	Can this works?
> b.	Any other information I need sync or restore other than information in database.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)