Windows 2003 Cluster not setting up 2nd node

Howdy-

In this two-node cluster, both servers can see the shared storage fine (mapped as Q:
drive, reads/writes from both), but when we follow the procedure in the 'Guide to
Configuring a Server Cluster Under Windows 2003', the first node sets up properly, but
when the 2nd node gets added, the drive goes offline for the 2nd node.

We're following the guide faithfully, but can't seem to determine what's wrong with our
setup...

TIA,
BM

"Brian MXP" <> wrote in message
news:...
> Howdy-
>
> In this two-node cluster, both servers can see the shared storage fine
(mapped as Q:
> drive, reads/writes from both), but when we follow the procedure in the
'Guide to
> Configuring a Server Cluster Under Windows 2003', the first node sets up
properly, but
> when the 2nd node gets added, the drive goes offline for the 2nd node.

Actually, this is correct, the disk is only online on 1 node at any given
time.
So if you correctly configured Node1, and the disk is on Node 1, then when
adding Node 2, the access to the disk is lost. This is by design. Microsoft
Failover Clustering is a "shared nothing" model, which means the Disks are
not shared by the nodes, instead they are online on 1 node.

You need to "move the group" in cluster administrator to Node 2, then you
can access the disk from Node 2 (but you cannot from Node 1)

Rgds,
Edwin.



>
> We're following the guide faithfully, but can't seem to determine what's
wrong with our
> setup...
>
> TIA,
> BM

Edwin,

Thank you for the quick reply.

Just to follow up, even if we ignore the second node no longer able to
read/write to the Quorum, adding the second node fails.

Has anyone tried to add the second node then gotten
"Error 0x800706be: The Remote procedure call failed"

This error is several places during the final configuration does not allow
you to move on in the process.

Thank you,
TK

"Edwin vMierlo [MVP]" wrote:

>
> "Brian MXP" <> wrote in message
> news:...
> > Howdy-
> >
> > In this two-node cluster, both servers can see the shared storage fine
> (mapped as Q:
> > drive, reads/writes from both), but when we follow the procedure in the
> 'Guide to
> > Configuring a Server Cluster Under Windows 2003', the first node sets up
> properly, but
> > when the 2nd node gets added, the drive goes offline for the 2nd node.
>
> Actually, this is correct, the disk is only online on 1 node at any given
> time.
> So if you correctly configured Node1, and the disk is on Node 1, then when
> adding Node 2, the access to the disk is lost. This is by design. Microsoft
> Failover Clustering is a "shared nothing" model, which means the Disks are
> not shared by the nodes, instead they are online on 1 node.
>
> You need to "move the group" in cluster administrator to Node 2, then you
> can access the disk from Node 2 (but you cannot from Node 1)
>
> Rgds,
> Edwin.
>
>
>
> >
> > We're following the guide faithfully, but can't seem to determine what's
> wrong with our
> > setup...
> >
> > TIA,
> > BM
>
>
>

This sounds like name resolution issues...triple-check all your name
resolution.

--
Ryan Sokolowski
MVP - Clustering
MCT, MCITP x3, MCTS x7, MCSE x2, CCNA, CCDA, BCFP


"Tina Kane" <Tina > wrote in message
news:E5012EA8-022F-4F0B-A66F-...
> Edwin,
>
> Thank you for the quick reply.
>
> Just to follow up, even if we ignore the second node no longer able to
> read/write to the Quorum, adding the second node fails.
>
> Has anyone tried to add the second node then gotten
> "Error 0x800706be: The Remote procedure call failed"
>
> This error is several places during the final configuration does not allow
> you to move on in the process.
>
> Thank you,
> TK
>
> "Edwin vMierlo [MVP]" wrote:
>
>>
>> "Brian MXP" <> wrote in message
>> news:...
>> > Howdy-
>> >
>> > In this two-node cluster, both servers can see the shared storage fine
>> (mapped as Q:
>> > drive, reads/writes from both), but when we follow the procedure in the
>> 'Guide to
>> > Configuring a Server Cluster Under Windows 2003', the first node sets
>> > up
>> properly, but
>> > when the 2nd node gets added, the drive goes offline for the 2nd node.
>>
>> Actually, this is correct, the disk is only online on 1 node at any given
>> time.
>> So if you correctly configured Node1, and the disk is on Node 1, then
>> when
>> adding Node 2, the access to the disk is lost. This is by design.
>> Microsoft
>> Failover Clustering is a "shared nothing" model, which means the Disks
>> are
>> not shared by the nodes, instead they are online on 1 node.
>>
>> You need to "move the group" in cluster administrator to Node 2, then you
>> can access the disk from Node 2 (but you cannot from Node 1)
>>
>> Rgds,
>> Edwin.
>>
>>
>>
>> >
>> > We're following the guide faithfully, but can't seem to determine
>> > what's
>> wrong with our
>> > setup...
>> >
>> > TIA,
>> > BM
>>
>>
>>

Maybe I was confused by the documentation, but when prompted during the
configuration of the first node by Cluster Services to use a uniqe cluster IP
address, I had created a third private IP (one private IP for each node and
one public IP for each node) - should I have used the private IP of the first
node instead of a third one?

If not, I can ping the private and public IP and hostnames from both nodes
including the third IP assigned to the cluster.

Thanks for any clarification.

TK

"Ryan Sokolowski [MVP]" wrote:

> This sounds like name resolution issues...triple-check all your name
> resolution.
>
> --
> Ryan Sokolowski
> MVP - Clustering
> MCT, MCITP x3, MCTS x7, MCSE x2, CCNA, CCDA, BCFP
>
>
> "Tina Kane" <Tina > wrote in message
> news:E5012EA8-022F-4F0B-A66F-...
> > Edwin,
> >
> > Thank you for the quick reply.
> >
> > Just to follow up, even if we ignore the second node no longer able to
> > read/write to the Quorum, adding the second node fails.
> >
> > Has anyone tried to add the second node then gotten
> > "Error 0x800706be: The Remote procedure call failed"
> >
> > This error is several places during the final configuration does not allow
> > you to move on in the process.
> >
> > Thank you,
> > TK
> >
> > "Edwin vMierlo [MVP]" wrote:
> >
> >>
> >> "Brian MXP" <> wrote in message
> >> news:...
> >> > Howdy-
> >> >
> >> > In this two-node cluster, both servers can see the shared storage fine
> >> (mapped as Q:
> >> > drive, reads/writes from both), but when we follow the procedure in the
> >> 'Guide to
> >> > Configuring a Server Cluster Under Windows 2003', the first node sets
> >> > up
> >> properly, but
> >> > when the 2nd node gets added, the drive goes offline for the 2nd node.
> >>
> >> Actually, this is correct, the disk is only online on 1 node at any given
> >> time.
> >> So if you correctly configured Node1, and the disk is on Node 1, then
> >> when
> >> adding Node 2, the access to the disk is lost. This is by design.
> >> Microsoft
> >> Failover Clustering is a "shared nothing" model, which means the Disks
> >> are
> >> not shared by the nodes, instead they are online on 1 node.
> >>
> >> You need to "move the group" in cluster administrator to Node 2, then you
> >> can access the disk from Node 2 (but you cannot from Node 1)
> >>
> >> Rgds,
> >> Edwin.
> >>
> >>
> >>
> >> >
> >> > We're following the guide faithfully, but can't seem to determine
> >> > what's
> >> wrong with our
> >> > setup...
> >> >
> >> > TIA,
> >> > BM
> >>
> >>
> >>
>

Maybe I'm still confused by the description, but you should basically have 2
IPs for the public NICs (1 per node), 2 more IPs in a different subnet for
the heartbeat NICs and one more IP for the Cluster itself. Additional IPs
would be needed for additional (read: actually functioning) services within
the cluster: MSDTC, SQL, Exchange, etc...

--
Ryan Sokolowski
MVP - Clustering
MCT, MCITP x3, MCTS x7, MCSE x2, CCNA, CCDA, BCFP


"Tina Kane" <> wrote in message
news12DDE68-6858-48FA-B369-...
> Maybe I was confused by the documentation, but when prompted during the
> configuration of the first node by Cluster Services to use a uniqe cluster
> IP
> address, I had created a third private IP (one private IP for each node
> and
> one public IP for each node) - should I have used the private IP of the
> first
> node instead of a third one?
>
> If not, I can ping the private and public IP and hostnames from both nodes
> including the third IP assigned to the cluster.
>
> Thanks for any clarification.
>
> TK
>
> "Ryan Sokolowski [MVP]" wrote:
>
>> This sounds like name resolution issues...triple-check all your name
>> resolution.
>>
>> --
>> Ryan Sokolowski
>> MVP - Clustering
>> MCT, MCITP x3, MCTS x7, MCSE x2, CCNA, CCDA, BCFP
>>
>>
>> "Tina Kane" <Tina > wrote in message
>> news:E5012EA8-022F-4F0B-A66F-...
>> > Edwin,
>> >
>> > Thank you for the quick reply.
>> >
>> > Just to follow up, even if we ignore the second node no longer able to
>> > read/write to the Quorum, adding the second node fails.
>> >
>> > Has anyone tried to add the second node then gotten
>> > "Error 0x800706be: The Remote procedure call failed"
>> >
>> > This error is several places during the final configuration does not
>> > allow
>> > you to move on in the process.
>> >
>> > Thank you,
>> > TK
>> >
>> > "Edwin vMierlo [MVP]" wrote:
>> >
>> >>
>> >> "Brian MXP" <> wrote in message
>> >> news:...
>> >> > Howdy-
>> >> >
>> >> > In this two-node cluster, both servers can see the shared storage
>> >> > fine
>> >> (mapped as Q:
>> >> > drive, reads/writes from both), but when we follow the procedure in
>> >> > the
>> >> 'Guide to
>> >> > Configuring a Server Cluster Under Windows 2003', the first node
>> >> > sets
>> >> > up
>> >> properly, but
>> >> > when the 2nd node gets added, the drive goes offline for the 2nd
>> >> > node.
>> >>
>> >> Actually, this is correct, the disk is only online on 1 node at any
>> >> given
>> >> time.
>> >> So if you correctly configured Node1, and the disk is on Node 1, then
>> >> when
>> >> adding Node 2, the access to the disk is lost. This is by design.
>> >> Microsoft
>> >> Failover Clustering is a "shared nothing" model, which means the Disks
>> >> are
>> >> not shared by the nodes, instead they are online on 1 node.
>> >>
>> >> You need to "move the group" in cluster administrator to Node 2, then
>> >> you
>> >> can access the disk from Node 2 (but you cannot from Node 1)
>> >>
>> >> Rgds,
>> >> Edwin.
>> >>
>> >>
>> >>
>> >> >
>> >> > We're following the guide faithfully, but can't seem to determine
>> >> > what's
>> >> wrong with our
>> >> > setup...
>> >> >
>> >> > TIA,
>> >> > BM
>> >>
>> >>
>> >>
>>

Ryan,

Sorry for the confusion, but you have the scenario down

I have two public IP's, (one for each node) and two private IP's (one for
each node - NetBIOS is disabled) - the two nodes can ping each other by IP's
and hostnames for both public and private.

I have another IP configured for the cluster itself and this is the UP and
NetBIOS name I used during the initial setup. It sounds like that is
accurate.

The hardware on both is the same and I'm runnning Win2003 R2 64bit EE.

I have the Quorum configured for both nodes, but I still get that RPC error.
I checked to verify my clusteradmin account has all the correct permissions
via local sec. pol as well as local admin group. All the services necessary
are started so I'm at a loss with all the thousands of clustering articles on
what that error means.

Any light you can shed would be terrific, and thanks again for the response.

TK

"Ryan Sokolowski [MVP]" wrote:

> Maybe I'm still confused by the description, but you should basically have 2
> IPs for the public NICs (1 per node), 2 more IPs in a different subnet for
> the heartbeat NICs and one more IP for the Cluster itself. Additional IPs
> would be needed for additional (read: actually functioning) services within
> the cluster: MSDTC, SQL, Exchange, etc...
>
> --
> Ryan Sokolowski
> MVP - Clustering
> MCT, MCITP x3, MCTS x7, MCSE x2, CCNA, CCDA, BCFP
>
>
> "Tina Kane" <> wrote in message
> news12DDE68-6858-48FA-B369-...
> > Maybe I was confused by the documentation, but when prompted during the
> > configuration of the first node by Cluster Services to use a uniqe cluster
> > IP
> > address, I had created a third private IP (one private IP for each node
> > and
> > one public IP for each node) - should I have used the private IP of the
> > first
> > node instead of a third one?
> >
> > If not, I can ping the private and public IP and hostnames from both nodes
> > including the third IP assigned to the cluster.
> >
> > Thanks for any clarification.
> >
> > TK
> >
> > "Ryan Sokolowski [MVP]" wrote:
> >
> >> This sounds like name resolution issues...triple-check all your name
> >> resolution.
> >>
> >> --
> >> Ryan Sokolowski
> >> MVP - Clustering
> >> MCT, MCITP x3, MCTS x7, MCSE x2, CCNA, CCDA, BCFP
> >>
> >>
> >> "Tina Kane" <Tina > wrote in message
> >> news:E5012EA8-022F-4F0B-A66F-...
> >> > Edwin,
> >> >
> >> > Thank you for the quick reply.
> >> >
> >> > Just to follow up, even if we ignore the second node no longer able to
> >> > read/write to the Quorum, adding the second node fails.
> >> >
> >> > Has anyone tried to add the second node then gotten
> >> > "Error 0x800706be: The Remote procedure call failed"
> >> >
> >> > This error is several places during the final configuration does not
> >> > allow
> >> > you to move on in the process.
> >> >
> >> > Thank you,
> >> > TK
> >> >
> >> > "Edwin vMierlo [MVP]" wrote:
> >> >
> >> >>
> >> >> "Brian MXP" <> wrote in message
> >> >> news:...
> >> >> > Howdy-
> >> >> >
> >> >> > In this two-node cluster, both servers can see the shared storage
> >> >> > fine
> >> >> (mapped as Q:
> >> >> > drive, reads/writes from both), but when we follow the procedure in
> >> >> > the
> >> >> 'Guide to
> >> >> > Configuring a Server Cluster Under Windows 2003', the first node
> >> >> > sets
> >> >> > up
> >> >> properly, but
> >> >> > when the 2nd node gets added, the drive goes offline for the 2nd
> >> >> > node.
> >> >>
> >> >> Actually, this is correct, the disk is only online on 1 node at any
> >> >> given
> >> >> time.
> >> >> So if you correctly configured Node1, and the disk is on Node 1, then
> >> >> when
> >> >> adding Node 2, the access to the disk is lost. This is by design.
> >> >> Microsoft
> >> >> Failover Clustering is a "shared nothing" model, which means the Disks
> >> >> are
> >> >> not shared by the nodes, instead they are online on 1 node.
> >> >>
> >> >> You need to "move the group" in cluster administrator to Node 2, then
> >> >> you
> >> >> can access the disk from Node 2 (but you cannot from Node 1)
> >> >>
> >> >> Rgds,
> >> >> Edwin.
> >> >>
> >> >>
> >> >>
> >> >> >
> >> >> > We're following the guide faithfully, but can't seem to determine
> >> >> > what's
> >> >> wrong with our
> >> >> > setup...
> >> >> >
> >> >> > TIA,
> >> >> > BM
> >> >>
> >> >>
> >> >>
> >>
>

Thanks to everyone that replied. It ended up that I had Jumbo MTU set to
9000 on the Private Vlan and needed to set that to 1500 - wacky!

TK

Not only that coz in our case the MTU settings were Ok on the servers and the
switches.
Turns out to be an issue with the Routing and Remote service, we disabled it
on both nodes and voilÃ* the second node was successfully added to the cluster
configuration.

BR,


"Tina Kane" wrote:

> Thanks to everyone that replied. It ended up that I had Jumbo MTU set to
> 9000 on the Private Vlan and needed to set that to 1500 - wacky!
>
> TK
>
>