May 29, 2008 at 9:56 am
I had made changes to enable AWE and updated the min and max server memory setting yesterday. This morning we failed over the cluster to the other node (active/passive cluster). Since the failover stops sql server service on node1 and starts it on node2, I thought that the configuration changes should have taken affect. But, they did not. Is this expected behavior, or are we possibly missing a step?
There were no error messages in the errorlog or event viewer indicating something was wrong. There was a message regarding AWE:
"Set AWE Enabled to 1 in the configuration parameters to allow use of more memory."
Windows 2003 R2, 32-bit
SQL Server 2005, 32-bit
Thank you in advance!
May 29, 2008 at 2:44 pm
Where are the system databases on this cluster, are they on a shared drive that both sets of SQL services are set to use, or are they residing locally. I know that VERITAS does not care, and if you do not set this up so the services start the master from the shared disk any changes will be done to the local mdf file. just a geus
Andrew
May 29, 2008 at 2:55 pm
andrewkane17 (5/29/2008)
Where are the system databases on this cluster, are they on a shared drive that both sets of SQL services are set to use, or are they residing locally. I know that VERITAS does not care, and if you do not set this up so the services start the master from the shared disk any changes will be done to the local mdf file. just a geusAndrew
Andrew - thank you for replying. Our system database files do reside on a shared drive.
The setting for AWE (along with other settings viewed by sp_configure) are stored in master.sys.configurations. After the failover occurred, the config_value for awe was still 1, so it knew it should be enabled. I assumed the failover would have taken care of it because sql server service stops on node1 and then starts on node2. Does the stopping and starting need to be done on the same physical machine? Has anyone else encountered this problem.
Another thought is that we did not reboot the machine after we updated the lock pages in memory initially -- maybe that had something to do with it?
May 29, 2008 at 2:59 pm
Was the lock pages in memory set on both nodes? I have set this up on many VERITAS clusters and have followed http://www.sqlservercentral.com/articles/Administering/aweadventures/960/ to do the steps, just wondering?
May 29, 2008 at 3:06 pm
andrewkane17 (5/29/2008)
Was the lock pages in memory set on both nodes? I have set this up on many VERITAS clusters and have followed http://www.sqlservercentral.com/articles/Administering/aweadventures/960/ to do the steps, just wondering?
Yes. I updated both nodes yesterday (without rebooting) and then the failover occurred.
Do I need to update lock pages in memory, REBOOT, and then enable awe, and restart sql server service (or reboot).
May 29, 2008 at 3:19 pm
Give it a reboot, can't hurt.
May 29, 2008 at 3:25 pm
andrewkane17 (5/29/2008)
Give it a reboot, can't hurt.
Thanks Andrew. We scheduled a reboot for tomorrow morning. I'm guessing that will do the trick. I will be sure to report back. I still would like to figure out what happened for future reference.
Viewing 7 posts - 1 through 6 (of 6 total)
You must be logged in to reply to this topic. Login to reply