Re: Cluster Failovers

Tech Tip: Click here to run a free scan for Windows Errors and optimize PC performance

John,
From what you said below:
"The preferred owners setting only affects the "Failback" behavior and group
failover behavior of 3+ node clusters."

My question is: can there be automatic failback in a two-node cluster? I
have been trying this without any success. Failover works like a charm, but
not failback. I am initiating failure by either stopping a service on one
node or unplugging the public network cable.

Thank you
RM

"John Toner [MVP]" wrote:

DV,

This is the intended behavior. Groups will "ping pong" between nodes until
the group threshold is breached...which is set to 10 by default.

The preferred owners setting only affects the "Failback" behavior and group
failover behavior of 3+ node clusters.

Regards,
John


"DV" <clubv@xxxxxxxxxxx> wrote in message
news:1148467083.822152.191400@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
Hi,

I have a quick question for any clustering experts out there. This
question relates to MSCS on Windows 2003..

I have been doing some quick labs testing failover policies and there
is a query that i have.

Ive set up a 2 node cluster called CLUSTER1 with nodes CLUSTERNODE1 and
CLUSTERNODE2 within the cluster.

I have setup a group called "Print Group" which holds a disk resource
and a spooler resource, just for the sake of testing the failover
policy.

I have setup CLUSTERNODE1 and my preffered owner for this group and
CLUSTERNODE1 now owns the "Print Group"

When i initiate a failure to one of the resources in the "Print Group",
the cluster fails the "Print Group" to CLUSTERNODE2. Great.. thats what
i expect...

But when i initiate another failure to the "Print Group" on
CLUSTERNODE2 it initiates a failover back to CLUSTERNODE1..

Im confused.. I would of though that MSCS would realise that there
was/is a fault with CLUSTERNODE1 hence the original failover of the
group to CLUSTERNODE2 and it would just fail the resource and leave the
"Print Group" where it is, but instead it does fail it back to the
original "faulty" node...

If i continue to fail the resources in the group, MSCS failsover
between the nodes until the failover threshold is met.. which is what i
except...

So my question is... what is the default behavour of MSCS? To failover
between the original and standby node continuesly until the failover
threshold is met? Does it not take into account that a failed Group has
come from a specific server and then remember not to failover to that
server should the group fail again?


Thanks for your time

DV.




.

Relevant Pages

  • Re: Cluster Failovers
    ... I have been doing some quick labs testing failover policies and there ... Ive set up a 2 node cluster called CLUSTER1 with nodes CLUSTERNODE1 and ... CLUSTERNODE1 now owns the "Print Group" ... but instead it does fail it back to the ...
    (microsoft.public.windows.server.clustering)
  • Re: Cluster Failovers
    ... Yes there can be automatic failback, but are you sure that you would want to ... failure that causes the group to failover from NodeA to NodeB. ... Ive set up a 2 node cluster called CLUSTER1 with nodes CLUSTERNODE1 ... I have setup a group called "Print Group" which holds a disk resource ...
    (microsoft.public.windows.server.clustering)
  • Cluster Failovers
    ... I have been doing some quick labs testing failover policies and there ... Ive set up a 2 node cluster called CLUSTER1 with nodes CLUSTERNODE1 and ... CLUSTERNODE1 now owns the "Print Group" ... I would of though that MSCS would realise that there ...
    (microsoft.public.windows.server.clustering)