Hi,
I have SQL 2005 server (with Named Instance) running on a two node Cluster
Configuration (Active/Passive).

SQL Instance, SQL Server Agent, SQL Server Browser services are running
under a service account (domain account). Similarly Cluster service is also
running under a cluster service account (domain account).

I am looking for accurate steps to update password of SQL Service account
and Cluster service account that will cause minimum disruption of these
services. If there’s a link to documentation on how to update password, that
will be most useful.

For the cluster service account password:
http://support.microsoft.com/kb/305813/en-us

For the SQL 2005 Instance, use the SQL Configuration Manager to change the
password. It handles the "cluster magic".

--
Geoff N. Hiten
Senior SQL Infrastructure Consultant
Microsoft SQL Server MVP




"Mehul" <Mehul (AT) discussions (DOT) microsoft.com> wrote

It could be just me, but I had problem using Configuration Manager to change
the SQL service account password from time to time. I don't remember the
exact error message now, but I remember not having success in getting the
change replicated among the nodes.

I've had more success with just using the services.msc mgmt console to
change the SQL service account password on each node. This is just a hassle
since on each node there are a few services to change. I know Configuration
Manager is not cluster aware, but had hoped that at least any change made
through it to the SQL registry entries would be automatically replicated by
the cluster service.

Linchi

"Geoff N. Hiten" wrote:

Thanks Geoff. But I would really like to know what are the steps to perform.
Do I need to change password on both the nodes or only on the active node ?
Once the password for the account is updated in AD, is it immediately
reflected on the SQL servers ?

"Geoff N. Hiten" wrote:

Linchi, Thanks for your feedback.

So what steps do you follow to update password for all these services using
services.msc ?

"Linchi Shea" wrote:

Here are the steps I went through recently to update the password for the SQL
service account (the previous password expired. It shouldn't be set to
expire, but that's a different story):

1. Remote desktop to each node
2. Start -> Run, type services.msc
3. In the service list, locate all the SQL Server related services that use
the password.
4. Double click on each such service
5. Click on the "Log On' tab.
6. Type in the new password in the Password and Confirm Password textboxes.
7. Click on Apply and OK.

Linchi

"Mehul" wrote:

LInchi's advice on using the services applet on each node is a bit more
complex, but it is a certaintly.

As for when changes "take", AD may need up to fifteen minutes to replicate
the password change through the system when you have multiple controllers.
If an account is logged in to a resource, such as a service account already
running, you can leave it running or a while before changing it. The system
will not force it out immediately but the application may no longer be able
to access network resources after some time.

--
Geoff N. Hiten
Senior SQL Infrastructure Consultant
Microsoft SQL Server MVP




"Mehul" <Mehul (AT) discussions (DOT) microsoft.com> wrote

Linchi,
Here's what worked for me. Quite similar to the steps you have outlined:

1. Changed SQL Service account password in AD. Waited for about half an hour
so that password change gets replicated to all the domain controllers.
2. Updated the password on the active node of the SQL Cluster using SQL
Configuration manager for all the services.
3. Restarted the services using SCM. Everything looked fine till now.
4. On the passive node, using Services MMC, manually updated password for
the SQL services.
5. Fail over the cluster from node 1 to node 2, everything worked fine with
no errors.

I will be blogging these steps, but hope that other people who are in the
same situation will find this discussion useful.

Thanks Geoff and Linchi for your quick responses.


"Linchi Shea" wrote:

I'm glad that you included step 5 to failover the SQL group among the nodes.
That's an abosolutely critical step to close loop the whole task.

Linchi

"Mehul" wrote: