May 12, 2005 at 7:41 am
I've got an Active/Passive cluster (sql server 2000) and occassionally the load will get split. Is there a notification service that I can enable so that I can receive an email when this event occurs?
May 12, 2005 at 7:48 am
Do you want to monitor SQL Server failover?
May 12, 2005 at 8:00 am
I just want to know when it gets in that condition so that I can get the entire load onto one machine. Would monitoring do that and send me an email?
Right now the only way I know that a split occurs is by me physically launching the cluster manager.
May 12, 2005 at 8:18 am
Could you be a little more specific about what you mean by a 'split cluster?'
If it is an Active/Passive cluster, that would mean that there is only one active SQL Server virtual server on the cluster (say it's running on Node A). Node B is standing by in case something happens to Node A and it needs to take over runnning the SQL Server virtual serer (a failover).
Windows clustering doesn't do load balancing (well, SQL Server clustering at least. You can do application clusters with IIS). If you have an Active/Active cluster, you are really running two SQL Server virtual servers, one on each node.
Long story short, there really isn't anything to 'split' between the clusters, if I am understanding your question correctly.
May 12, 2005 at 8:50 am
I think I'm asking my question wrong, sorry. I'm a programmer tasked with administering a cluster until the position gets filled. How can I get a notification when a failover occurs?
What I'm seeing is that Node A fails over and then the cluster administrator --> cluster group and sql are still indicating that some resources in the Cluster Group are still assigned to the machine that failed over.
At that point I manually move the load over and restart the machine that failed, then move the load back after a restart.
May 12, 2005 at 9:48 am
You can write a small script by using the cluster command to find out on what resource your cluster is running and schedule the same.
e.g.
cluster /cluster:server_name group "disk group 1" > node.txt
read the output of node.txt and see if you can find the servername and you can take actions appropriately.
May 13, 2005 at 10:00 am
Without knowing anything specific about your cluster setup, I would suspect that you have a configuration problem with the cluster. The cluster groups should be set up with all of the resources in the group failing over as a unit. If the group fails over, you should get SQL Server, SQL Server Agent, the ipaddress, network name, and all disks going at the same time.
Are there any additional cluster resources in the SQL Server cluster group that aren't necessary (defined as a dependency in the SQL Server cluster resource) in the cluster group? Sometimes people add a disk for scratch space or backups, etc. that isn't set up as a dependency, so it won't necessarily fail over along with a SQL Server virtual server. If this is the case you would have to decide between adding it as a dependany to SQL (adding another failure point) or trying to build a script or something to make it failover right after the SQL Server failed over.
I don't know if the stuff above was of any help, but if not, post more specifics and i'll see if I can make a specific suggestion.
Chris
May 13, 2005 at 1:25 pm
Chris -
The data and backup drives are listed under sql in the cluster administrator but not in the cluster group... the quorum is listed in the cluster group
I did notice that the quorum drive name is not the same between the 2 machines the drive letter is though.
Viewing 8 posts - 1 through 7 (of 7 total)
You must be logged in to reply to this topic. Login to reply