[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

RE: [Linux-cluster] MySQL Failover / Failback



Well, if you prevent mysql from starting on boot, your load balancer should recognize that and not fail back to that server. We took mysql itself out of the cluster since we also do replication. It was not necessary to have it in there for our application.
 
Rob Gil
Linux Systems Administrator
American Home Mortgage
 


From: linux-cluster-bounces redhat com [mailto:linux-cluster-bounces redhat com] On Behalf Of Bjorn Oglefjorn
Sent: Thursday, July 26, 2007 4:33 PM
To: linux clustering
Subject: Re: [Linux-cluster] MySQL Failover / Failback

Thanks Rob, but that's not quite what I'm looking for.  I have a middleware script of my own which allows me to have both nodes with mysql running at all times.  I need to ensure that when the MASTER fails, Cluster Suite will failover to the SLAVE.  My script then performs some SQL ops to allow writes, etc.  When the MASTER comes back from being fenced, I need to make sure that Cluster Suite does not failback to the MASTER.  I'm hoping that the 'nofailback' option exists and will work as I expect it to.

--BO

On 7/26/07, Robert Gil <Robert Gil americanhm com> wrote:
I had figured this out a while back and forgot to post back to the list.
The steps are as follows

Create a child script to the virtual ip to check whether mysql is up.
Have mysql NOT start on boot.

With these two things done. When the vip fails over, it will not fail
back because mysql wont be up.

Rob Gil
Linux Systems Administrator
American Home Mortgage

________________________________

From: linux-cluster-bounces redhat com
[mailto:linux-cluster-bounces redhat com] On Behalf Of Bjorn Oglefjorn
Sent: Thursday, July 26, 2007 3:18 PM
To: linux clustering
Subject: Re: [Linux-cluster] MySQL Failover / Failback


I found that a 'nofailback' option was added for the <failoverdomains>
section of the conf.  I can't find any reference to 'nofailback' in any
RHCS doc I can find.  I'm guessing it should look like this:

    <failoverdomain name="test_failover_domain" ordered="1"
restricted="1" nofailback="1">
        ...
    </failoverdomain>

Can someone confirm?  I will attempt to confirm this myself and will
report back when I know for sure.

--BO


On 7/26/07, Bjorn Oglefjorn <sys mailing gmail com> wrote:

        I'd love to know how to do this as well.  Anyone?
        --BO



        On 6/1/07, Robert Gil <Robert Gil americanhm com
<mailto:Robert Gil americanhm com > > wrote:

                1) Node 1 (master) dies
                        -How do we enable "sticky" failover so that it
does not then fail back to Node 1



--
Linux-cluster mailing list
Linux-cluster redhat com
https://www.redhat.com/mailman/listinfo/linux-cluster


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]