"Unable to connect" from the second node after a r

fmisto · Wed Oct 15, 2008 9:36 am

Hello,
I hope that I don't miss my request in the FAQ.

I explain my issue.
From a VMWARE server (free version 1.0.3), I installed two windows 2008 server (64 bit version) to build a cluster (for a test).
I follow step by step your documentation "Starwind isci targer for microsoft Windows:Using starwind with microsoft cluster on windows server 2008" ( firstly, it's a superb documentation).
I installed the starwind software (version 3.5.5 trial version 64 bit) on the first server (node 1) .
The second server (node 2) can join the iscsi target.(with firewall disable on the both server)
All was perfect, but on the page 67 of 96, the documentation requires to shut down the server node 2.
Since I did it, the server node 2 is unable to connect on iscsi target.
the message in "iscsi initiator" is "reconnecting".
I remove the iscsi target from the iscsi initiator, and tried to repeat the steps of page 55 to 61 without success.
I disconnect on server 1 the iscsi target on server 1, I removed the starwind device on the server 1, and repeat again all process to create starwind device on server 1.
From the server 1, I can connect the starwind device from iscsi initiator, but impossible from server 2. (the firewall are devalided on the both server).
Have you an idea about this issue?
could you tell me where I should have a look?

Thanks for your support
regards
Fabien

aaron (staff) · Wed Oct 15, 2008 10:37 am

Did you apply free key with only one connection allowed?

fmisto · Wed Oct 15, 2008 10:48 am

Hi,
No, I didn't.

regards
Fabien

Wed Oct 15, 2008 3:30 pm

Just great... OK, in such a case please zip and send us StarWind logs to support@rocketdivision.com so we could take a closer look at your issue. Thank you for cooperation!

Val (staff) · Wed Oct 15, 2008 10:10 pm

Hi!

1) To build a 2-way cluster you need the iSCSI target to run on a third machine, not a cluster node!

2) Check it whether you enabled clustering for the shared targets.

fmisto · Tue Oct 21, 2008 12:06 pm

OK,
I will do it.
please, Precise this point in your documentation.
Thanks for your clarifications
Regards
Fabien

Tue Oct 21, 2008 7:01 pm

Well... You've been the first one who was building HA configuration using the same machine for failing node and iSCSI target at the same time

fmisto wrote:OK,
I will do it.
please, Precise this point in your documentation.
Thanks for your clarifications
Regards
Fabien

Thona · Wed Oct 22, 2008 8:27 am

That may not be so unusual.

Naturally it makes zero sense from a HA point of view.

But as lab equpment, for teaching people how to set up a cluster, install it, maintain it, demonstrate failovers. You can even simulate hardware failure, but just turning a VM off.

That said, I would still do that with a minimum of 4 virtual machines.
* 2 for the cluster
* 1 for the domain controller
* one holding the ISCSI target.

May be three, if the host acts as ISCSI target itself.

Wed Oct 22, 2008 8:34 pm

Exactly. Zero sense. That's why I would not "clone" such a thing showing it to the students. My $0.02.

Thona wrote:That may not be so unusual.

Naturally it makes zero sense from a HA point of view.

But as lab equpment, for teaching people how to set up a cluster, install it, maintain it, demonstrate failovers. You can even simulate hardware failure, but just turning a VM off.

That said, I would still do that with a minimum of 4 virtual machines.
* 2 for the cluster
* 1 for the domain controller
* one holding the ISCSI target.

May be three, if the host acts as ISCSI target itself.