users lose connection during shared failover

[Paul_621]

When I manually move a group containing a share from one node to the other, the users lose connection to the share, but after a minute can reconnect - is this normal. Is this to be expected from a DL380G4 MSA500G@2 cluster with 2003 enterprise ? I was hopeing that the users would not see anything during a node failover, is this possible.

cheers

Paul

[Stephen Grimmer]

yes that is normal

[Doug de Werd]

yes this is normal. There is a network feature in Windows called "auto reconnect". What is does is to disconnect and reconnect clients from the network when there is no network activity - this happens all the time during normal operations. So the cluster takes advantage of this during a failover - the connection is lost during the "move" butu the auto-reconnect feature will conenct the clients again after a short time.

Youc an test this further by watching a client that is not doing any network activity during the move. They should see no disruption at all. Usually you will only get an error message on the client side if you are doing network reads or writes at the time of the move (or failover).

Here is more info on it:

http://support.microsoft.com/default.aspx?scid=kb;en-us;297684

Thanks,
Doug

[Tv_3]

Paul this is an expected behavior during a faillover...

You can prevent from failing back (and loosing another couple of minutes)during the bussiness hours, you are able to shedule to failbackup at specific time.

[Paul_621]

Thanks for all the advice, it would seem that users losing connection during failover is a normal thing. Which is what I suspected, but the funny thing is, I contacted HP australia - many minutes on the phone, and they reassured me that no one should lose their connections during a failover ?? SO I rebuilt the Cluster to their recommended specs install 2003 enterprise - no SP1, PSP 7.30 and repeat for second server - create new cluster. But the same thing happen again, users losing connection during failover. hmm - I guess the real world answer prevails here.

Many thanks for the replies.

cheers

Paul

[Tv_3]

You have altimes downtime during a failover / failback.

For a sql server it can take a few minutes before the surveying node takes over all roles. Seen it on a few large data base environments, contected microsoft for it and the pointed me to a document stating this issue unfortunaly i can remember the link.

Sincerly.

I see your a newbie to the forums, once again welcome.

Please to do not forgot to close the thread and please thake a minute to assign points for those who took some time to aswer you.
Thank you!

[Paul_621]

Thanks everybody, as a new person to this forum this was very helpful.


cheers


Paul