MMM for DR Site

[RdbmsDBA]

Hi Guys,

I am in the process of setting up mmm for our solution.  My setup is that I have 2 masters on primary (only one writable with VIP) and 2 slaves to the master 1 on DR site.
This is all up and running and no issues with it.

In addition, I want to setup a MMM server on DR site when/if we have to do a failover.  This mmm will run in passive mode.
I have installed the server and have defined the configuration.  In this setup, my 2 databases on DR site will be my active masters and I am not sure if I make 2 primary databases as slaves for this configuration(essentially we will only be switching to DR if both production servers are not available).  The main problem that I see is that my slaves on DR site in this configuration needs to be made as masters and start replicating from each other in the case of failover.

This means, when a failover happens, DBA will have to start replication between these 2 masters (previous slaves of the primary) and the change mmm server to active mode.  Can somebody help if they have got similar setup or is there a better solution for this?  Once proudction is available again, take a backup from DR (current production), restore on production and start mmm on production again.

The biggest problem is how to test this setup.  When I have to test failover, I will have to rebuild all masters etc. 

Has some kind soul got some test cases/programs etc to test the mmm setup in primary and dr site etc.  Will be a huge help.
Many thanks,
Rick

[Walter Heck]

I would probably opt for a little less complicated solution:

A and B are master pair on your primary site,

on secondary site, you have C which replicates from B and D replicating from C. This means that when you have a complete fail of your primary site, your secondary site is already up and running. Failover cul even happen automatically. Obviously this means no HA for the secondary site, but if that runs only in emergency situations, it should  never be your main site for very long and you might be willing to take that risk, reducing a lot of complexity and making it ossible to automate very step of what I described.

Hope this helps,

cheers,

Walter

--
You received this message because you are subscribed to the Google Groups "MySQL Multi Master Manager Development" group.
To view this discussion on the web visit https://groups.google.com/d/msg/mmm-devel/-/AUWwOykuU5MJ.
To post to this group, send email to mmm-...@googlegroups.com.
To unsubscribe from this group, send email to mmm-devel+...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/mmm-devel?hl=en.

--
Walter Heck

--
follow @walterheck on twitter to see what I'm up to!
--
Check out my new startup: Server Monitoring as a Service @ http://tribily.com
Follow @tribily on Twitter and/or 'Like' our Facebook page at http://www.facebook.com/tribily

[RdbmsDBA]

Hey Walter, thanks for taking out time to reply.  I am slightly confused by you reply.

A and B are master pair on your primary site, Right

on secondary site, you have C which replicates from B and D replicating from C.

If servers on Primary are A and B, Secondary are C and D. 

With my current setup, I have B,C and D all replicating from A.  Is it ok or should I change it and why?

This means that when you have a complete fail of your primary site, your secondary site is already up and running. Failover cul even happen automatically. Obviously this means no HA for the secondary site, but if that runs only in emergency situations, it should  never be your main site for very long and you might be willing to take that risk, reducing a lot of complexity and making it ossible to automate very step of what I described.

Are you suggesting no mmm for secondary site at all?  For failover, we need a manual failover right?  If this can be automated, how?  We need to restart slaves on Secondary as master.  I think, its pretty simple to have mmm for secondary that just controls the master and don't bother about slaves (previous primary).

Has anybody got similar setup?  Is it feasible/working?  Can somebody share their thoughts and if possible, Test cases and testing strategy?

Many thanks