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

Re: [Linux-cluster] rhel6 node start causes power on of the other one


On 3/25/11, Gianluca Cecchi <gianluca cecchi gmail com> wrote:
> I was referring to possible effects when problems arise.

Let us cross the bridge when we come to it. See answers below.

> I meant that with an approach similar to the ServiceGuard's one, you
> can have for example a condition where both nodes reboot (such as
> mutual stonith) and if there is a problem with intra cluster
> communication that persists during the immediate reboot, both nodes
> would remain at the prompt asking for confirmation.

On which CPU  should service guard run on?

An RHCS is HA Cluster (HPC is a different RH product and HA offerings
like MRG is another different RH product and RHEV is a very diferent
one altogether in Rehat as I understand -- did not have right hardware
to get RHEV going though, as yet) is designed to be self managed ( I
mean, without the "head node" which is present in a HPC cluster)

I am talking about Only RHCS here.

I can't comment more at this point of time about the other products
that you or I are talking about.

> So if you don't explicitly go on the console of one of them your
> "cluster services" remain down indefinitely.

Of course. It will. When you have no legs, you can't run.

> Instead, in rhcs, if for example the problem is resolved say after 5
> minutes, probably there would be 2 or 3 mutual stoniths and at the 4th
> attempt (when problem solved) they would form the cluster again and
> give service...

Please see above.

Perhaps I have not understood RHCS enough.

Above IMHO, of course.



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