Can't set up a MSCS two nodes cluster.

[georgeyu100]

Setup an iSCSI server By Starwind 5.4 on Win7, and two Win 2003 VMs as cluster nodes on vSphere 4.0. Using the Starwind way to setup Quorum and another disk shared between 2 VMs. The way is that on each VM by Microsoft's iSCSI initiator to get the two shared disks directly from iSCSI Server (win 7). When setup a cluster on any node, get the info on picture 1

cluster 1.JPG (31.08 KiB) Viewed 15428 times

. At the end of setup, get the error info: The network path was not found (picture 2)

cluster 4.JPG (34.93 KiB) Viewed 15426 times

, checked Win system logs, get the info on picture 3

cluster 5.JPG (141.23 KiB) Viewed 15426 times

. Before and after it, check the 3 NICs' IP and ping them, all of them are good. After put new IP Address/subnet mask on the NICs which are for heartbeat, the problem is still there.

I'm confused what is wrong. Is it a problem that the heartbeat line connects through a vSwitch without physical NIC as Microsoft said heartbeat line should be connect directly

[Constantin (staff)]

According to MS KB you have choosen IP for cluster that already is assigned to another NIC.
Link to KB http://technet.microsoft.com/en-us/libr ... S.10).aspx

[georgeyu100]

According to MS KB you have choosen IP for cluster that already is assigned to another NIC.
Link to KB http://technet.microsoft.com/en-us/libr ... S.10).aspx

Thanks. I got it!
But now, there is almost same problem at the end of configuration. After click the buttom Finish, get the the error message, please look picture 1. clock OK on it, get almost same error message. And, on systen log, there is a DCOM error. Please look picture 3.

My IP address:

10.1.1.11/24 on node1 for public network; 10.1.2.11/24 on node1 for iSCSI; 10.1.3.11/24 on node1 for private network.
10.1.1.22/24 on node2 for public network; 10.1.2.22/24 on node2 for iSCSI; 10.1.3.22/24 on node2 for private network.
10.1.3.10/24 for Cluster.
So, I don't know what is error.
But, I also get another info which is different with MS's MSCS document. please look picture 2. The difference is that on MSCS document, here are 4 Marks. But, you see, I get only 1 Mark.
Is it a problem?
My environment is like that vSphere 4.0 with 2 VMs which are on Win2003 SP2 and Microsoft iSCSI initiator as 2 nodes of MSCS. The private network NICs connects each other by a vSwitch without physical NIC. So, the infrustructure is following document StarWind_MSCluster2003.PDF. only the difference is that both nodes are not physical machines, they are VMs.
As MS documant said: private NICs (heartbeat) can't connect throught a routing device because the TTL=1 Sec was too long.
So, is it a problem?

[georgeyu100]

Sorry, I also tried public IP address 10.1.1.10 here, got same error message.
thanks!

[mooseracing]

As MS documant said: private NICs (heartbeat) can't connect throught a routing device because the TTL=1 Sec was too long.
So, is it a problem?

On server 2008 I had to do direct connect for heartbeat or my clusters kept thinking the partner was dead. It would randomly bring the cluster up and down.

[Max (staff)]

I am sorry, maybe I've got something wrong but in the header you've mentioned a Win7 and 2 2003 boxes.
What does HA heartbeat do here? or is this a Node-node cluster heartbeat meant?

[georgeyu100]

I am sorry, maybe I've got something wrong but in the header you've mentioned a Win7 and 2 2003 boxes.
What does HA heartbeat do here? or is this a Node-node cluster heartbeat meant?

Yes, Starwind 5.4.0.1579 on Win 7 64bit, 2 nodes on Win2003 SP2 64bit on vSphere 4.0 host. heartbeat between 2 nodes connects by a vSwitch. it is a node-node cluster.

Sorry, I can't understand your question: What does HA heartbeat do here?

Thanks!

[anton (staff)]

Start with updating to 5.5 as an initial step. With HA heartbeat Max asked about StarWind HA heartbeat, not MS cluster heartbeat. As h/b was represented with 5.5 and up you can ignore this question.

[georgeyu100]

Start with updating to 5.5 as an initial step. With HA heartbeat Max asked about StarWind HA heartbeat, not MS cluster heartbeat. As h/b was represented with 5.5 and up you can ignore this question.

Thanks for your reply.
Do you mean 5.5 no this problem? and Starwind has its conception called Starwind HA heartbeat on 5.5? In fact, I tried 5.4 for VMware HA successfully before. Could you give me a link which is an explain or documant for Starwind HA heartbeat ?

Thank again.

George

[georgeyu100]

Now, I fixed the proble. But, after added second node, got another error message:
The quorum disk could not be located by the cluster service.please look the picture 1.
I checked the Starwind, the disk created on Clustered. Please looked the Picture 2.
and Computer Managerment on Win2003 on both nodes. Please looked picture 3.
Every thing lokked fine.

What is the problem?
Thankks!

[Max (staff)]

Did you doublecheck that the drive is connected to both servers?

[georgeyu100]

Did you doublecheck that the drive is connected to both servers?

Yes, I can write a file down on the disk from both nodes.

[Max (staff)]

This is kind of a problem which usually misteriously disappears after Windows reinstall, but this is the last option.
I will dig the technet forums for more info on this case, please stay tuned!

[georgeyu100]

This is kind of a problem which usually misteriously disappears after Windows reinstall, but this is the last option.
I will dig the technet forums for more info on this case, please stay tuned!

I rebooted them 2-3 time.
Thank you for your help!

[Max (staff)]

Unfortunately no new info from me on this time, George, can you please update us about your progress?