Page 1 of 1

Server 2003 hard shutdown halts cluster

Posted: Wed Aug 13, 2008 12:05 pm
by tkloppel
First, we're using the trial for Starwind version 3.5.5 to evaluate the product.

Created a 2GB quorum image which is configured with default options and enabled for multiple connections (clustering.)

I have created the cluster without any issues, and it appears to function properly.

I can shut down a node and the other takes over without delay, I can initiate failures and move the cluster group between nodes. But when I pull the power cord on node1, node2 is not able to take over. It attempts 3 times and then kills the cluster service. In the Starwind log, it looks like there is a lock or some kind of reservation on the quorum target from node1 that is preventing node2 from taking control.

Once the cluster service gets restarted, node2 becomes active -- but this is not until after about 3 minutes.

Any help here would be great. I can send logs if needed.

Posted: Thu Aug 14, 2008 5:12 pm
by anton (staff)
1) Are you using the most recent version of StarWind?

2) If "yes" to 1) please send the logs to support@rocketdivision.com (zipped if you please). + detailed description what you do :)

Thank you very much for cooperation!

Re: Server 2003 hard shutdown halts cluster

Posted: Wed Sep 24, 2008 5:43 pm
by Shawn
tkloppel wrote:First, we're using the trial for Starwind version 3.5.5 to evaluate the product.

Created a 2GB quorum image which is configured with default options and enabled for multiple connections (clustering.)

I have created the cluster without any issues, and it appears to function properly.

I can shut down a node and the other takes over without delay, I can initiate failures and move the cluster group between nodes. But when I pull the power cord on node1, node2 is not able to take over. It attempts 3 times and then kills the cluster service. In the Starwind log, it looks like there is a lock or some kind of reservation on the quorum target from node1 that is preventing node2 from taking control.

Once the cluster service gets restarted, node2 becomes active -- but this is not until after about 3 minutes.

Any help here would be great. I can send logs if needed.
I am having the same problem with the evaluation software in an almost identical setup. I am running the trial version which I downloaded about a week ago:

StarWind iSCSI Target v3.5.4 (Build 20080527, Win64)
Trial license, days left: 24)

I even seem to be seeing the same log messages as mentioned above ("ScsiOp 0x28 - Reservation conflict.") However my node 2 never becomes active and eventually stops trying to attach. I did notice in the StarWind device properties that Persistent reservation is set to "Yes". I am not sure if this can be turned off or is the problem.
Has a resolution been proposed or found?

Posted: Thu Sep 25, 2008 9:54 pm
by Val (staff)
Try to change the following parameter in StarWind.cfg and restart the service:
from <iScsiPingPeriod value="0"/>
to <iScsiPingPeriod value="5"/>

This should help.

Posted: Mon Sep 29, 2008 11:03 pm
by Shawn
valery (staff) wrote:Try to change the following parameter in StarWind.cfg and restart the service:
from <iScsiPingPeriod value="0"/>
to <iScsiPingPeriod value="5"/>

This should help.

It worked!

Posted: Tue Sep 30, 2008 6:59 am
by anton (staff)
Good :)