Failover settings (newbie)

After reading the documentation I'm still confused on what the Windows 2008
failover cluster settings mean. Please assume a 2-node Node and Disk Majority
cluster.

Maximum failures in the specified period.
What happens after the maximum is reached within the specified period of
time? Does the cluster and its services stay down?

Regardless of the max failures and the time period, does the cluster still
try to failover to a secondary node when the primary node fails? Or does it
first try to restart on the primary node? If it first tries to restart on
primary, how many times does it try to restart on primary before it fails
over to secondary?

What does the setting "Prevent failback" mean? If some time after failing
over to the secondary node, the secondary node fails, will it fail over back
to the primary node? Or will the cluster just go down?

Why would you set it to Allow failback immediately? Don't you need some time
to diagnose or do other work (e.g., run updates) on the node that has gone
down?

If there are two clustered services (in my case, MSDTC and SQL Server),
shouldn't the failover settings be identical for both?

When you specify Move this service or application to another node, does that
simulate failover?

Thanks for your help.

"Nicole G." <Nicole G.@discussions.microsoft.com> wrote in message
news:A53C7298-2835-4544-9942-...

> Maximum failures in the specified period.
> What happens after the maximum is reached within the specified period of
> time? Does the cluster and its services stay down?

There are two settings. One setting is a property of the resource itself,
and the other setting is a property of the cluster group.

The setting for the resource causes a failure, and the failure causes the
resource to automatically recover on the same node as the failure. Once the
threshold is reached, then the failure of the resource will cause the entire
cluster group to failover to the other node, assuming that the affect group
check box is enabled.

After the failures (failover to another node) exceed the cluster group
setting, the entire cluster group will go offline in a failed state.

> Regardless of the max failures and the time period, does the cluster still
> try to failover to a secondary node when the primary node fails?

If the entire node failes, yes, unless it has reached the threshold for the
cluster group in which case it will fail and go offline.

> What does the setting "Prevent failback" mean?

This means if the cluster group failed over to the other node, it would fail
back to the original node as soon as the orginal node is available again.
With prevent failback, the cluster group will stay on the other node when it
fails over until another failure or a manual process moves the cluster group
back.

> Why would you set it to Allow failback immediately? Don't you need some
> time
> to diagnose or do other work (e.g., run updates) on the node that has gone
> down?

Yes, you want to do that, which is why almost nobody ever sets it to allow
failback immediately.

> If there are two clustered services (in my case, MSDTC and SQL Server),
> shouldn't the failover settings be identical for both?

It depends on too many different issues to cover here.

> When you specify Move this service or application to another node, does
> that
> simulate failover?

No, it doesn't simulate failover, it is just a move. If you want to simulate
a fail over, fail one of the resources multiple times until you reach the
threshold.

--
Russ Kaufmann,
MVP, MCSE: Messaging and Security, MCT, MCITP, MCTS and other stuff

ClusterHelp.com, a Microsoft Certified Gold Partner
Web http://www.clusterhelp.com
Blog http://msmvps.com/clusterhelp

"Russ Kaufmann" <> wrote in message
news:...
> "Nicole G." <Nicole G.@discussions.microsoft.com> wrote in message
> news:A53C7298-2835-4544-9942-...
>
>> Maximum failures in the specified period.
>> What happens after the maximum is reached within the specified period of
>> time? Does the cluster and its services stay down?
>
> There are two settings. One setting is a property of the resource itself,
> and the other setting is a property of the cluster group.
>
> The setting for the resource causes a failure,

Sorry, this does not read right. Obviously the setting doesn't cause a
failure. The setting is used to define behaviour during a failure.