SQL Server Performance Forum – Threads Archive
Cluster group dependenciesI am installing sql 2005 active/passive cluster over windows 2003 SP2. The clustering service wasn’t installed by me. When I started to test the cluster before I did SQL installation, I found there was only 1 cluster group available. I created a new group ‘SQL Server Group’ and when I tried to move disk array to the new group it shows all the resources along with my disks will be moved to the new group. How do I only move disk arrays and not the rest of the resources (quorum drive, ip address and cluster name, etc)? Why is this working so? Some mis-configuration in dependencies? What should be dependent on what? I need to start my sql installation ASAP. Any help is greatly appreciated.
Welcome to SSP forum… I don’t think it is misconfigured…because these resources are dependent on each other… Check the following… Failover cluster resource dependencies in SQL Server
SQL-Server-Performance.com All postings are provided â€œAS ISâ€ with no warranties for accuracy.
[V]Thanks but none of those help me. The person who installed cluster tells me that when he did the cluster installation, the 2nd node was not turned off. After the installation he say all the disk drives has come up in different groups (Cluster group with IPAddress, Network Name and Quorum, Group 0 with L, Group 1 with M, Group 2 with N, Group 3 with N and Group 4 with S). He then proceeded to change all the disk groups to Cluster Group. Later on I created ‘SQL Server Group’ and moved all except quorum to this new group. Now, when I failover/move this new group, the cluster group stays on node 1 and only sql server group moves to node 2. Shouldn’t all the resources move to the second node? How do I fix this? Arti
It has been my experience that when you begin to have any trouble with the installation of a cluster that you need to start over from scratch, instead of trying to fix the problem. I suggest you uninstall Windows clustering and reinstall it according to the suggestions on this website. Be sure that the shared array has a drive designated as a Quorum drive and at least on drive for data (if not more). Be sure to test the server to see that if fails over properly before you begin to install SQL Server Clustering. If it won’t work now, it won’t work after SQL Server is installed. ——————————–
Brad M. McGehee, SQL Server MVP
Thank you very much for your help. The management doesn’t agree to uninstallation though I have explained the whole issue in detail. They have asked me to continue if test for ‘Manually failover nodes by Turning them off’ works fine. They only foresee this kind of situation in real life senario (1 box goes down at a time). Is it absolutely necessary to have seperate groups? For me to continue to install SQL 2005, currently I have moved all the resources to 1 group – Cluster Group (bcs when all resources are in 1 group they still failover together to the secondary node). MSDTC is also in the same group. Anybody foresee any issue if I install SQL 2005 (SP2) on this setup? Thanks!
The best practice is to have two separate resource groups, although it is possible to only have one. One group will work, but has limitations. It is possible that this current configuration might be fixed, but I would have to see exactly how it was configured. Doing this over a forum is not practical. I don’t think adding SP2 will add to your problems. But real answer is what I explained before. It appears that the original clustering was not properly done. If you can’t get your management to believe you, have them contact me, and for an expensive consulting fee, I’ll tell them the same thing. Maybe if an expensive outside consultant says you need to rebuild the Windows clustering, they will believe me. I feel your distress in this issue, as managment often does fully understand the technology they "supposedly" manage. For example, there is no downside to uninstalling and then reinstalling Windows clustering, so doing so is a no brainer. ——————————–
Brad M. McGehee, SQL Server MVP
[<img src=’/community/emoticons/emotion-1.gif’ alt=’‘ />]<br />Thank you. I will pass on the message. [<img src=’/community/emoticons/emotion-5.gif’ alt=’‘ />]<br /><br />Arti
Actually, it doesn’t sound like it needs to be rebuilt from the ground up…<br /><br />1st — Move the SQL resources out of the Cluster Resource Group and into thier own resource group dedicated to the first instance on the cluster.<br /><br />2nd — Create a secondary SQL Resource Group for the second instance on the cluster. This will require you to create additional disk, ip and network name resources.<br /><br />3rd — Ensure that your Cluster Resource Group only has resources in it dedicated to running the cluster (no SQL resources).<br /><br />4th — Understand that cluster resource groups can reside on different nodes – that’s how active/active works in the first place.<br /><br />I’d be happy to help you sort out this issue without rebuilding, for a cost of course. <img src=’/community/emoticons/emotion-1.gif’ alt=’‘ /><br /><br />Send me a PM and we can work out the details if you wish.