Clustering failover loosing client connections

  • Thread starter Thread starter Jerry
  • Start date Start date
J

Jerry

Running Win2k Adv server with clustering if the cluster
is failed over users loose there connection to Notes,
Word, Excel. Anyone have any idea as to why this is
happening?
 
It's the way cluster works. This is by design.

When failover occurs, it's the same as stopping the service and restarting
it. So, for instance, the server service stops on node A and starts on node
B. No state is transferred when the resources fail over, so clients must
re-establish their connections.

It's always been that way with cluster.

-Matt
 
Can you provide more details on what is happening and what you are expecting
?
Are you running those apps(Word, Excel) as a clustered resources ?

If I understand your question correctly, you will loose the connection but
you will be able to reconnect once the concerned resources come online.

Thanks,
Sandeep Sutari
Microsoft Corp.

This posting is provided "AS IS" with no warranties, and confers no rights.
Use of any included script samples are subject to the terms specified at
http://www.microsoft.com/info/cpyright.htm
 
My Clustered servers are running active active and are
for file services only.

They are set up as follows:

Server1 holds the Users home directories and IT data
This would be the users personal MS documents and Lotus
Notes email configuration files and database files.

Server2 hold the Departmental files and Applications.

If something happens to Server1 and the resources fail
over to Server2 or visversa the users should not loose
their work or be knocked off the system. This is the
whole purpose of Clustering is to stay up and not cause
down time.

The apps are loaded onto the workstations so they cannot
be setup as clustered resources.

If the drives are shared out as a resource and it fails
from one server to the other why whould the clients fail?

Regards,

Jerry Talbot
(e-mail address removed)


Our clients will
 
Hi Jerry,
This is kind of a common misconception of clustering. Clusting is not fault
tolerant it is highly available. Fault tolerance allows for seesion state
information to be saved and moved, high availability does not. It should
also be noted that this is not an active-active scenario, active-active
means the same group or data set is running on both nodes. The scenario
described is an active-passive/load-balanced scenario; where each node has
an active group that can fail to the other node.

When a group fails all session data is lost (this is true for all things on
clusters), essentially a failover should be conceptualized as the server
completly failing and a new server coming on with the same name. This new
server has no concept of the session state of the first server. Some
application overcome this by having a cache (SQL for example) where
uncommited changes are kept and stored on client machines or in shared
storage. When they fail-over the application knows to look for these files
and then do that work first before taking any new work. In this way a
majority, but not all, of the working data at the time of failure can be
recovered.

The bottom line is this is just how cluster works at this time, when a
cluster resource fails all session data is lost and the clients essentialy
must start a new session with a new server (even though its the same
virtual server).
--
Curtis Koenig
Support Engineer
Product Support Services, Security Team
MCSE, MCSES, CISSP

This posting is provided "AS IS" with no warranties and confers no rights.
Please reply to the newsgroup so that others may benefit. Thanks!

--------------------
 
Thanks Curtis. Unfortunately this is not how our MS rep
explained it to us when we decided to setup the cluster.
We wanted a system that if we had a problem could switch
it over whatever the time may be and the users would not
see any issues.

So are there any documents on how this situation would
have to be configured to work to make it fault tolerant?
Thanks again for your help in this matter.
 
Hi Jerry,
Its not possible to make file shares fault tolerant (or really anything for
that matter). This is really a technical limitation of clustering as the
core technology is not desiged to be fault tolerant. As I stated before
some applications have designed themselves to overcome this somewhat, but
its not part of the cluster core and therefore there is no design that will
make this work as desired in this situation.
--
Curtis Koenig
Support Engineer
Product Support Services, Security Team
MCSE, MCSES, CISSP

This posting is provided "AS IS" with no warranties and confers no rights.
Please reply to the newsgroup so that others may benefit. Thanks!

--------------------
 
Back
Top