May 23, 2003 at 3:05 am
Hi,
I'm fairly new to clustering. I want to switch on a couple of trace flags in a particular instance in our cluster. Will stopping the instance cause any problems ? I'm just a bit nervous that it might cause a failover or something.
Thanks.
Growing old is mandatory, growing up is optional
May 23, 2003 at 4:59 am
When you say Stopping the instance it means stopping the SQL service using the service applet. or using the SQL server Manager. Thsi is not advisable. The Cluster Administrator monitors the status of these services and once its stopped he will start it. He does it the no of times specified in the SQL Resource property. By default it restarts the service 3 times. The fouth time it will failover. In order to stop SQL server its better to take the SQL resources offline using the Cluster Administator.
Hope this clears your doubt
Jesus My Saviour
May 23, 2003 at 5:50 am
Many Thanks, it does.
Growing old is mandatory, growing up is optional
May 23, 2003 at 7:26 am
Not completely true. I have a 2-node active/passive cluster with a virtual SQL Server. I have stopped services several times and never experienced a failover FROM stopping the services MANUALLY. When the services have stopped due to outside events (not manually stopped), the cluster immediately failed over.
-SQLBill
May 23, 2003 at 7:52 am
Best practice would still be to stop and start resources through Cluster Administrator.
K. Brian Kelley
http://www.truthsolutions.com/
Author: Start to Finish Guide to SQL Server Performance Monitoring
http://www.netimpress.com/shop/product.asp?ProductID=NI-SQL1
K. Brian Kelley
@kbriankelley
May 23, 2003 at 8:09 am
You can take the resources offline and then stop the services. The status of the node is checked every 0.5 secs and it responds within 0.2 secs. If it is failed, heartbeats are sent for about 5.3 secs at regular intervals before physically failing over the resource. If the re-start process falls within this time, it won't failover.
May 27, 2003 at 6:32 am
While it is true you can stop the service manually (without going through cluster services), there is the possibliity that the node will lock up, hold all the disk resouces, and prevent the other node from coming on-line (this happened to me one weekend when I was doing remote work and carelessly shut the service down with EM). The only solution was a hands-on power-down to free up the resources from the hosed node.
May 27, 2003 at 7:52 am
This is what experts have to say:
Tip: Don't turn off or on the mssqlserver, sqlserveragent, or the msdtc services off or on using any other tool other than Cluster Administrator when SQL Server 7.0 has been clustered. That means don't use Enterprise Manager, Service Manager, or Services to turn these services off or on.
Explanation: If you don't follow this rule, you could corrupt your SQL Server 7.0 cluster installation and have to reinstall it.
Also keep in mind that sometimes SQL Server itself, will prompt you to stop or start one of the services. For example, when you change a server setting that requires you to stop or start a service for it to be in effect, SQL Server often asks if you want to stop and then restart the appropriate service at this time. If you are asked this, always say no, and then use Cluster Administrator to stop and start the appropriate service.
In addition, be sure that "Autostart SQL Server," "Autostart SQL Server Agent," and "Autostart MSDTC" are all unselected in the "General" tab of the SQL Server "Properties" screen, and that "Auto restart SQL Server if it stops unexpectedly" and "Auto restart SQL Server Agent if it stops unexpectedly" are both unselected in the "Advanced" tab of the SQL Server Agent "Properties" screen.
In SQL Server 2000, the mssqlserver, sqlserveragent, or the msdtc services can be stopped and started from Enterprise Manager and Service Manager.
From : http://www.sql-server-performance.com/clustering.asp
Franco
Franco
May 29, 2003 at 7:49 am
Yes, Always use the cluster admin to stop or start any of the SQL Services. Never start or stop the services from EM or services. I restarted SQL Agent through services a few times and there was no problem but one day I did it and it failed over to node 2. From that point on I always use the cluster admin.
Viewing 9 posts - 1 through 8 (of 8 total)
You must be logged in to reply to this topic. Login to reply