Cluster failover causes
-
Maybe someone knows some links where I (and I believe others too) may find some valuable information about main and others causes of cluster failovers and following actions what to do then? What to do and what not to prevent from it. It's not a problem for us yet but we wanna be prepared to this. We have active\active sql 2000 on win2003 servers, all work fine but there is nothing eternal
I hope you understood me and what I would like to get
For any help thanks in advance
-
In real world scenarios with clusters pretty much the only occassion I've seen a cluster failover automatically was when one of the nodes crashed due to a hardware fault.
Other than that you are only really looking at disk problems (related to hardware controllers, scsi cables, hba troubles), or if SQL itself crashes for some reason.
-
Thanks Nicholas. You have mentioned about physical or hardware problems here. What's about the software problems? Have you ever experenced or maybe heard about it? I mean maybe there are some issues that work on simple SQL server but doesn't work on clustered SQL server. or something liek that For inst.: when our programmers wanted to put one programm on the sql server cluster it didn't work here, but it worked fine on a simple SQL server.
Interesting what are the most critical places on the cluster...
Simple it's hard to believe that the cluster is like a find out of all our problem
I hope you understood me
-
You will only get the software related failover if the application built is cluster aware or runs on a service that you can manage through the cluster software.
As an example I'm using an application that connects to another server for notifications, this is not a cluster aware piece of software and needs to be running on the same server that the sql instance is running on. This service sometimes crashes, so I have it set up in the same group as SQL and have set it to affect the group and failover when it fails.
There are software related issues and support issues when it comes to clustering, for example, Microsoft says that SQL Mail is not supported in a clustered environment, although it does work (most of the time). You would have to test a particular application to see it it will function correctly within the cluster.
-
Hi,
an other issue is SQLServerAgent...
If is crashes due to job-related problems, do you really want to automagically failover (and eventually disconnect your users?
Do you have other software which needs to use the same disks as sql server and therefore resides in the same cluster-group? perhaps your some backup software?
And of course one reason for failover may be the admin, if he is stopping services manually...
regards
karl
Best regards
karl -
You can remove any services (ie sql server agent, full text search) from forcing a failover on service failure, so that's not so much of a problem.
-
Yes, of course you can do that...
it is just something you have to think about:
if you do logshipping and the agent fails, you perhaps should failover, if it's just reporting you do, i would not want to switch nodes...
Best regards
karl
Viewing 7 posts - 1 through 6 (of 6 total)
You must be logged in to reply to this topic. Login to reply