Hi everybody,

due to a busy server (the 2nd node), I couldn't install the two nodes in
one step, and I was forced to install a cluster with one node, and bring it
online with the production database. Now everything works fine (in one node
of course) and I'd like to install the second node. I've just installed
windows 2003 sp1 in the 2nd node server, but i'm not sure the correct steps
to add a second node with the first one in production.

If you have any information, I'll really apreciate it.

Thanks

Patrick.

BOL covers exactly this topic under "Maintaining a Failover Cluster"

Index search "Failover Clustering". Choose maintaining. the first topic on
teh page is "Adding a Node to an existing Virtual Server".

Geoff N. Hiten
Microsoft SQL Server MVP

"Patrick Mac Kay" <pmackay_at_hotmail.com> wrote

The only risk of data corruption is before the LUNs are tagged as cluster
disks. Until then, both nodes will try and mount them, typically with bad
results. Once the cluster service writes an intentionally mangled signature
on the disks, the OS thinks they are foreign partitions and won't try and
mount them directly. You will need to do whatever your SAN vendor/software
requires so the new node can see the LUNs as well as the old node. You will
likely have to reboot the new node so it can see the LUNs as disks. Then
you can add the node with the cluster wizard (here is where Windows 2003
makes it a lot easier) and finally add the SQL instance per BOL. Don't
forget to apply any service packs and hotfixes to the new node.

Geoff N. Hiten
Microsoft SQL Server MVP

"Patrick Mac Kay" <pmackay_at_hotmail.com> wrote

Geoff,

thanks. It seems very simple, but I have some doubts. Is there any risk
with the databases or the shared directories in the SAN?. I read some
information about the clusters and the posibility of corrupt data if both
are online at the same time.... is that true?

Thanks again.

Patrick.

"Geoff N. Hiten" <sqlcraftsman (AT) gmail (DOT) com> escribió en el mensaje
news:ew7kcrwWFHA.2796 (AT) TK2MSFTNGP09 (DOT) phx.gbl...

thanks again. I'll try it today.....

Patrick.

"Geoff N. Hiten" <sqlcraftsman (AT) gmail (DOT) com> escribió en el mensaje
news:eNLG5ZxWFHA.2288 (AT) TK2MSFTNGP14 (DOT) phx.gbl...

Geoff,

almost everything works fine. The only problem I had was the path of the
sqlserver service excecutable, i.e., the path of the file sqlserver.exe. In
the second node I received the "access is denied" error when I tried to
start the service. That was because the path was configured as
"\\SQL02\c$\program~1\....." in the service, and the user who was configured
to start the service cannot access that path. I changed the user to a more
powerful user and it works great, but I don't like that solution. I think I
can change the path in the registry, and use a less power user... what to
you think about that?

Patrick.

"Patrick Mac Kay" <pmackay_at_hotmail.com> escribió en el mensaje
news:OOg2eeyWFHA.616 (AT) TK2MSFTNGP12 (DOT) phx.gbl...

On a cluster, the SQL service account, the cluster service account, and the
install account must be domain users in the local administrators group on
all nodes.

Geoff N. Hiten
Microsoft SQL Server MVP


"Patrick Mac Kay" <pmackay_at_hotmail.com> wrote

Patrick, I'm wondering if you got your second node setup correctly, and what
steps you took to get it working correctly.

I have basically the same situation, except that we're running 2000 Advanced
Server with a SCSI attached storage array. I followed the steps on the
following URL, but each time I did this the cluster service on node 2 hung in
the starting state, ending in an eventual hard reset of the server (ouch).

http://www.windowscluster.com/msdocs/stepbystep.htm

At this point I'd rather hire an expert who's done this a bunch, so I'm not
tinkering with a production system, but I'm having trouble finding a local
consultant (Orange County California) that I know is qualified and knows what
they're doing.

Our system has been like this for over a year, as I inherited it from the
knucklehead that didn't set it up correctly in the first place.

--
Patrick Rouse
Microsoft MVP - Terminal Server
http://www.workthin.com


"Geoff N. Hiten" wrote:

SCSI clusters can hang on adding or removing a node. Typically, the new
controller doesn't know it is peer node on a cluster so it tries to be the
only initiator on the bus. When it enumerates the SCSI devices, it hangs
the bus. Also, it is a very common mistake to forget to change the SCSI ID
of the second controller. That will hang the bus as well.

Geoff N.Hiten
Microsoft SQL Server MVP

"Patrick Rouse" <PatrickRouse (AT) discussions (DOT) microsoft.com> wrote