Cluster 2008 R2 with iSCSI

Hardware and OS:
2 server (Node1 and Node2) with Windows 2008 R2 Enterprise,
3 network adapters on each server configured as follows:
INT LAN (10.10.10.1/24, 10.10.10.2/24)
Heartbeat (1.1.1.1/30, 1.1.1.2/30)
iSCSI (192.168.0.1/24, 192.168.0.2/24)
1 SAN iSCSI MSA2312i with IP 192.168.0.3/24

Purpose:
Create a cluster with iSCSI SAN

I haven't other servers, then the two nodes are also domain controllers.

I installed 2008; I created the domain; I have enabled servers with the
iSCSI initiator and connected them to the SAN; I have formatted the two Lun
(Lun0 = Quorum; Lun1 = Data); I added on the servers, the Failover Cluster
feature; I ran The Cluster Validation Wizard, with zero failure and warning.

At this point I created the cluster with ip 10.10.10.3/24, assigned as the
quorum Lun0; configure the network cards :

INT (LAN client and cluster traffic)
Heartbeat (traffic cluster)
iSCSI (no traffic for cluster and client)

Now I want to test failover, so I disable the INT LAN card on Node1 .

From a client PC I noticed that the ping has stopped and then is restarted.
In the management of the cluster, the quorum has moved on Node2.
Until here everything ok. (right?)

Now I have re-enable INT LAN card on Node1.
I disabled the LAN card INT on Node2.

From a client PC I noticed that the ping has stopped and is not restarted.
In the management of the cluster, the quorum is still ONLINE on Node2, but
the quorum is unreachable.

If I re-enable the INT LAN on Node2, the cluster is still not working.
The only way to bring life the cluster, I have stop and start the service on
cluster Node2.

How can I fix?
Thank you.

p.s: sorry for my english

"Mino" <> wrote in message
news:2880D2DF-DCAC-4410-ADC4-...

> Hardware and OS:
> 2 server (Node1 and Node2) with Windows 2008 R2 Enterprise,
> 3 network adapters on each server configured as follows:
> INT LAN (10.10.10.1/24, 10.10.10.2/24)
> Heartbeat (1.1.1.1/30, 1.1.1.2/30)
> iSCSI (192.168.0.1/24, 192.168.0.2/24)
> 1 SAN iSCSI MSA2312i with IP 192.168.0.3/24
>
> Purpose:
> Create a cluster with iSCSI SAN
>
> I haven't other servers, then the two nodes are also domain controllers.
>
> I installed 2008; I created the domain; I have enabled servers with the
> iSCSI initiator and connected them to the SAN; I have formatted the two
> Lun
> (Lun0 = Quorum; Lun1 = Data); I added on the servers, the Failover Cluster
> feature; I ran The Cluster Validation Wizard, with zero failure and
> warning.
>
> At this point I created the cluster with ip 10.10.10.3/24, assigned as the
> quorum Lun0; configure the network cards :
>
> INT (LAN client and cluster traffic)
> Heartbeat (traffic cluster)
> iSCSI (no traffic for cluster and client)
>
> Now I want to test failover, so I disable the INT LAN card on Node1 .
>
> From a client PC I noticed that the ping has stopped and then is
> restarted.
> In the management of the cluster, the quorum has moved on Node2.
> Until here everything ok. (right?)
>
> Now I have re-enable INT LAN card on Node1.
> I disabled the LAN card INT on Node2.
>
> From a client PC I noticed that the ping has stopped and is not restarted.
> In the management of the cluster, the quorum is still ONLINE on Node2, but
> the quorum is unreachable.
>
> If I re-enable the INT LAN on Node2, the cluster is still not working.
> The only way to bring life the cluster, I have stop and start the service
> on
> cluster Node2.
>
> How can I fix?
> Thank you.

In my very poor opinion you are using too many networks.
I have quite a similar configuration but I'm using only
one network.
And when I unplug the network cable from Node1, suddently
Node2 comes up, and when I reconnect Node1 it suddently
appears as the passive Node1, and so on.

Bruno

"Bruno Campanini" <> wrote in message
news:...
> In my very poor opinion you are using too many networks.
> I have quite a similar configuration but I'm using only
> one network.
> And when I unplug the network cable from Node1, suddently
> Node2 comes up, and when I reconnect Node1 it suddently
> appears as the passive Node1, and so on.

Three is the right number in this case.

Something to remember, when the public network is disconnected from both
nodes, the cluster service will stop and all clustered services and apps.
When disconnecting the private network, the intracluster communications will
use the public NIC since it is in mixed mode.


--
Russ Kaufmann
MVP, MCT, MCITP x7, MCTS x9, MCSE x4, CTT+
ClusterHelp.com, a Microsoft Gold Certified Partner

Email:
http://www.clusterhelp.com
Blog: http://msmvps.com/clusterhelp

Hi Mino,

I would also use at minimum 3 NICs for your scenario :-)

I see that you generally had configured your cluster correctly and the
quorum does also his right "job". The cluster service tries to keep the
majority of votes in your case at minimum 2. Did you get an event 1177 on
Node1/2 -> http://technet.microsoft.com/en-us/l...50(WS.10).aspx ?
can you create cluster logs (cluster /log gen) and review them for any
ERR -> http://technet.microsoft.com/en-us/l...42(WS.10).aspx ?

here are some add. links about quorum in Windows 2008 :
Failover Cluster Step-by-Step Guide: Configuring the Quorum in a Failover
Cluster
http://technet.microsoft.com/en-us/l...20(WS.10).aspx

Appendix A: Details of How Quorum Works in a Failover Cluster
http://technet.microsoft.com/en-us/l...49(WS.10).aspx

Regards
Ramazan

"Mino" <> wrote in message
news:2880D2DF-DCAC-4410-ADC4-...
> Hardware and OS:
> 2 server (Node1 and Node2) with Windows 2008 R2 Enterprise,
> 3 network adapters on each server configured as follows:
> INT LAN (10.10.10.1/24, 10.10.10.2/24)
> Heartbeat (1.1.1.1/30, 1.1.1.2/30)
> iSCSI (192.168.0.1/24, 192.168.0.2/24)
> 1 SAN iSCSI MSA2312i with IP 192.168.0.3/24
>
> Purpose:
> Create a cluster with iSCSI SAN
>
> I haven't other servers, then the two nodes are also domain controllers.
>
> I installed 2008; I created the domain; I have enabled servers with the
> iSCSI initiator and connected them to the SAN; I have formatted the two
> Lun
> (Lun0 = Quorum; Lun1 = Data); I added on the servers, the Failover Cluster
> feature; I ran The Cluster Validation Wizard, with zero failure and
> warning.
>
> At this point I created the cluster with ip 10.10.10.3/24, assigned as the
> quorum Lun0; configure the network cards :
>
> INT (LAN client and cluster traffic)
> Heartbeat (traffic cluster)
> iSCSI (no traffic for cluster and client)
>
> Now I want to test failover, so I disable the INT LAN card on Node1 .
>
> From a client PC I noticed that the ping has stopped and then is
> restarted.
> In the management of the cluster, the quorum has moved on Node2.
> Until here everything ok. (right?)
>
> Now I have re-enable INT LAN card on Node1.
> I disabled the LAN card INT on Node2.
>
> From a client PC I noticed that the ping has stopped and is not restarted.
> In the management of the cluster, the quorum is still ONLINE on Node2, but
> the quorum is unreachable.
>
> If I re-enable the INT LAN on Node2, the cluster is still not working.
> The only way to bring life the cluster, I have stop and start the service
> on
> cluster Node2.
>
> How can I fix?
> Thank you.
>
> p.s: sorry for my english

> I haven't other servers, then the two nodes are also domain controllers.
Upps - had forgotton your DC configuration :-)

I'm unsure if this is really supported/works cause I personally haven't
seen/tested an configuration like that yet.
Possibly this can also have further impact on your testings results and I
would try to find out an good old server where which you can use for AD and
DNS - otherwise you possibly get a wrong user experience here.

.....but this shouldn't be the root cause here :-(

Regards
Ramazan


"Mino" <> wrote in message
news:2880D2DF-DCAC-4410-ADC4-...
> Hardware and OS:
> 2 server (Node1 and Node2) with Windows 2008 R2 Enterprise,
> 3 network adapters on each server configured as follows:
> INT LAN (10.10.10.1/24, 10.10.10.2/24)
> Heartbeat (1.1.1.1/30, 1.1.1.2/30)
> iSCSI (192.168.0.1/24, 192.168.0.2/24)
> 1 SAN iSCSI MSA2312i with IP 192.168.0.3/24
>
> Purpose:
> Create a cluster with iSCSI SAN
>
> I haven't other servers, then the two nodes are also domain controllers.
>
> I installed 2008; I created the domain; I have enabled servers with the
> iSCSI initiator and connected them to the SAN; I have formatted the two
> Lun
> (Lun0 = Quorum; Lun1 = Data); I added on the servers, the Failover Cluster
> feature; I ran The Cluster Validation Wizard, with zero failure and
> warning.
>
> At this point I created the cluster with ip 10.10.10.3/24, assigned as the
> quorum Lun0; configure the network cards :
>
> INT (LAN client and cluster traffic)
> Heartbeat (traffic cluster)
> iSCSI (no traffic for cluster and client)
>
> Now I want to test failover, so I disable the INT LAN card on Node1 .
>
> From a client PC I noticed that the ping has stopped and then is
> restarted.
> In the management of the cluster, the quorum has moved on Node2.
> Until here everything ok. (right?)
>
> Now I have re-enable INT LAN card on Node1.
> I disabled the LAN card INT on Node2.
>
> From a client PC I noticed that the ping has stopped and is not restarted.
> In the management of the cluster, the quorum is still ONLINE on Node2, but
> the quorum is unreachable.
>
> If I re-enable the INT LAN on Node2, the cluster is still not working.
> The only way to bring life the cluster, I have stop and start the service
> on
> cluster Node2.
>
> How can I fix?
> Thank you.
>
> p.s: sorry for my english

Updates:

I did the test in a virtual environment, with 2 member servers and 1 DC,
same result, so even if the cluster is created with 2 dc there is no
difference.

I added a resource to the cluster with Lun1 and ip 10.10.10.4/24.
I dropped time Fail to 1 hour (the default is 6 hours).

I did the same tests.
I have disabled INT LAN on Node1 and the cluster is passed on Node2.
I have re-enabled the INT LAN on Node1, I waited 1 hour and I disabled the
INT LAN on Node2. The cluster is passed on Node1.

At this point I said, it works ok finally.... mah ...... I got the bad idea
to do another test.

Instead of disabling the card INT LAN servers, because hardly anybody
disable the network card but it is more likely that is missing the link to
the network, I tried unplugging the network cables in LAN INT network card of
the servers.

On Node1 I unplugged the cable from the rj45 network adapter INT LAN and
unfortunately the cluster is not passed on Node2.

Why does this happen? For the cluster is not the same thing to have the
network card disabled or missing links of the network card?

"Russ Kaufmann" wrote:

> "Bruno Campanini" <> wrote in message
> news:...
> > In my very poor opinion you are using too many networks.
> > I have quite a similar configuration but I'm using only
> > one network.
> > And when I unplug the network cable from Node1, suddently
> > Node2 comes up, and when I reconnect Node1 it suddently
> > appears as the passive Node1, and so on.
>
> Three is the right number in this case.
>
> Something to remember, when the public network is disconnected from both
> nodes, the cluster service will stop and all clustered services and apps.
> When disconnecting the private network, the intracluster communications will
> use the public NIC since it is in mixed mode.
>
>
> --
> Russ Kaufmann
> MVP, MCT, MCITP x7, MCTS x9, MCSE x4, CTT+
> ClusterHelp.com, a Microsoft Gold Certified Partner
>
> Email:
> http://www.clusterhelp.com
> Blog: http://msmvps.com/clusterhelp
>
> .
>

I noticed that the network cable from the ex-Node1 in Failover Cluster Manger
-> Networks -> Cluster Network 1 (my LAN INT) both cards are Fail.

Why?

"Mino" wrote:

> Updates:
>
> I did the test in a virtual environment, with 2 member servers and 1 DC,
> same result, so even if the cluster is created with 2 dc there is no
> difference.
>
> I added a resource to the cluster with Lun1 and ip 10.10.10.4/24.
> I dropped time Fail to 1 hour (the default is 6 hours).
>
> I did the same tests.
> I have disabled INT LAN on Node1 and the cluster is passed on Node2.
> I have re-enabled the INT LAN on Node1, I waited 1 hour and I disabled the
> INT LAN on Node2. The cluster is passed on Node1.
>
> At this point I said, it works ok finally.... mah ...... I got the bad idea
> to do another test.
>
> Instead of disabling the card INT LAN servers, because hardly anybody
> disable the network card but it is more likely that is missing the link to
> the network, I tried unplugging the network cables in LAN INT network card of
> the servers.
>
> On Node1 I unplugged the cable from the rj45 network adapter INT LAN and
> unfortunately the cluster is not passed on Node2.
>
> Why does this happen? For the cluster is not the same thing to have the
> network card disabled or missing links of the network card?
>
> "Russ Kaufmann" wrote:
>
> > "Bruno Campanini" <> wrote in message
> > news:...
> > > In my very poor opinion you are using too many networks.
> > > I have quite a similar configuration but I'm using only
> > > one network.
> > > And when I unplug the network cable from Node1, suddently
> > > Node2 comes up, and when I reconnect Node1 it suddently
> > > appears as the passive Node1, and so on.
> >
> > Three is the right number in this case.
> >
> > Something to remember, when the public network is disconnected from both
> > nodes, the cluster service will stop and all clustered services and apps.
> > When disconnecting the private network, the intracluster communications will
> > use the public NIC since it is in mixed mode.
> >
> >
> > --
> > Russ Kaufmann
> > MVP, MCT, MCITP x7, MCTS x9, MCSE x4, CTT+
> > ClusterHelp.com, a Microsoft Gold Certified Partner
> >
> > Email:
> > http://www.clusterhelp.com
> > Blog: http://msmvps.com/clusterhelp
> >
> > .
> >

"Russ Kaufmann" <> wrote in message
news:...
> "Bruno Campanini" <> wrote in message
> news:...
>> In my very poor opinion you are using too many networks.
>> I have quite a similar configuration but I'm using only
>> one network.
>> And when I unplug the network cable from Node1, suddently
>> Node2 comes up, and when I reconnect Node1 it suddently
>> appears as the passive Node1, and so on.
>
> Three is the right number in this case.

Well, have you tried to disconnect power supply cable
from active node?

Bruno

Third update:

Unfortunately, simulating the same situation in a virtual environment, I
found what's the problem.

Unplugging the network cable from Node1 goes to fail the handout of Node2 in
Failover Cluster Manager, as the servers are both Domain Controller. (So the
cluster with only 2 servers both DC, one can not ... at least for now)

Now I try to find a solution also hope in your help.

Thank you.

"RCan" wrote:

> > I haven't other servers, then the two nodes are also domain controllers.
> Upps - had forgotton your DC configuration :-)
>
> I'm unsure if this is really supported/works cause I personally haven't
> seen/tested an configuration like that yet.
> Possibly this can also have further impact on your testings results and I
> would try to find out an good old server where which you can use for AD and
> DNS - otherwise you possibly get a wrong user experience here.
>
> ....but this shouldn't be the root cause here :-(
>
> Regards
> Ramazan
>
>
> "Mino" <> wrote in message
> news:2880D2DF-DCAC-4410-ADC4-...
> > Hardware and OS:
> > 2 server (Node1 and Node2) with Windows 2008 R2 Enterprise,
> > 3 network adapters on each server configured as follows:
> > INT LAN (10.10.10.1/24, 10.10.10.2/24)
> > Heartbeat (1.1.1.1/30, 1.1.1.2/30)
> > iSCSI (192.168.0.1/24, 192.168.0.2/24)
> > 1 SAN iSCSI MSA2312i with IP 192.168.0.3/24
> >
> > Purpose:
> > Create a cluster with iSCSI SAN
> >
> > I haven't other servers, then the two nodes are also domain controllers.
> >
> > I installed 2008; I created the domain; I have enabled servers with the
> > iSCSI initiator and connected them to the SAN; I have formatted the two
> > Lun
> > (Lun0 = Quorum; Lun1 = Data); I added on the servers, the Failover Cluster
> > feature; I ran The Cluster Validation Wizard, with zero failure and
> > warning.
> >
> > At this point I created the cluster with ip 10.10.10.3/24, assigned as the
> > quorum Lun0; configure the network cards :
> >
> > INT (LAN client and cluster traffic)
> > Heartbeat (traffic cluster)
> > iSCSI (no traffic for cluster and client)
> >
> > Now I want to test failover, so I disable the INT LAN card on Node1 .
> >
> > From a client PC I noticed that the ping has stopped and then is
> > restarted.
> > In the management of the cluster, the quorum has moved on Node2.
> > Until here everything ok. (right?)
> >
> > Now I have re-enable INT LAN card on Node1.
> > I disabled the LAN card INT on Node2.
> >
> > From a client PC I noticed that the ping has stopped and is not restarted.
> > In the management of the cluster, the quorum is still ONLINE on Node2, but
> > the quorum is unreachable.
> >
> > If I re-enable the INT LAN on Node2, the cluster is still not working.
> > The only way to bring life the cluster, I have stop and start the service
> > on
> > cluster Node2.
> >
> > How can I fix?
> > Thank you.
> >
> > p.s: sorry for my english
>

"Mino" <> wrote in message
news:6087EA64-3098-470A-ADF2-...
> Third update:
>
> Unfortunately, simulating the same situation in a virtual environment, I
> found what's the problem.
>
> Unplugging the network cable from Node1 goes to fail the handout of Node2
> in
> Failover Cluster Manager, as the servers are both Domain Controller. (So
> the
> cluster with only 2 servers both DC, one can not ... at least for now)

Again, both nodes being DCs is not a good idea by any means.

http://msmvps.com/blogs/clusterhelp/...-bad-idea.aspx
http://support.microsoft.com/default.aspx/kb/281662
http://support.microsoft.com/kb/295091
http://support.microsoft.com/kb/898634
http://support.microsoft.com/kb/171390

--
Russ Kaufmann
MVP, MCT, MCITP x7, MCTS x9, MCSE x4, CTT+
ClusterHelp.com, a Microsoft Gold Certified Partner

Email:
http://www.clusterhelp.com
Blog: http://msmvps.com/clusterhelp