Hello,

Wondering if anyone can assist with this problem. We have
SQL 2000 clustered over two servers. SQL runs fine when on
server 01, but when we try to fail-over SQL to server 02,
the MSSQLServer logs the following errors to the NT event
log. All have MSSQLSERVER as the source, and all are event
ID 17052:

[sqsrvres] StartResourceService: StartService
(MSSQLSERVER) failed. Error: 42d

[sqsrvres] OnlineThread: ResUtilsStartResourceService
failed (status 42d)

[sqsrvres] OnlineThread: Error 42d bringing resource
online.

These three error messages are logged another 12 times
each (36 messages in total) over the next 12 minutes, and
then the SQL Server cluster resource is marked as Failed.

I cannot find any information about these errors on
Google, this newsgroup, or Microsoft Support. Any
assistance would be appreciated.

Regards,

Steve

Hi Steve -
The status info that you got with this error indicates
[sqsrvres] OnlineThread: ResUtilsStartResourceService
failed (status 42d). Here 42d is expressed in Hexadecimal. The decimal
equivalent is 1069.

From the command prompt, c:\> net helpmsg 1069 --> shows "The service did
not start due to a logon failure".

I would advise you to check the service startup account for both SQL and
SQL agent services. Recently if you had changed the password of this
windows account, then change the password in the service windows as well.

Ensure that the service startup account is same in both the nodes of your
cluster.

Let me know once you check this.

Regards,
Chandra


--------------------

Hi Chandra,

Thank you for your advice. I re-entered the correct
password for the SQL Server account in the Log On tab of
the MSSQLSERVER service, and I was then able to use Move
Group in Cluster Administrator to successfully move SQL to
either of the servers.

HOWEVER: That all worked yesterday; just out of
curiousity, today I used Cluster Administrator|Move Group
to move SQL to the other server again, but I am getting
the same error messages all over again.

It seems like something is killing the password of the SQL
Server account on the second server every so often. Yet I
am still able to log into the server using the SQL Server
user login and password. Very strange. Any ideas how I
could troubleshoot this further?

Regards,

Steve

Hi Steve -
Have you done this for both the nodes ? The error
message clearly indicates the login problem you have in your cluster.
Please check the following:

1) Check for SQL server service and change the password if required on both
the nodes.
2) Check for SQL agent service and change the password for agent as well on
both the nodes.
3) Now try bringing the sql group online and move the group to other nodes.

Once done, check the event log. Last time we got the error with status 42d.
Check whether the status is the same after doing the above 3 steps.

Did you at any time had any interconnect network problem between these 2
nodes? If so, it could be related to registry entries problem.

Are you having this problem on both the nodes or just consistently on one
node?
What is the OS user logged in to the nodes? Is this the same user or
different ones?

Let me know.

Regards,
Chandra



--------------------

All,

I have solved the problem. What was happening was every
night, on the two cluster servers the Group Policies were
being updated from the domain (I noticed some SceCli
messages in the NT eveng log saying "Security policy in
the Group policy objects are applied successfully." -
after these messages, SQL would fail when moving to
another server). And for some reason, in our Group Policy,
the user account that SQL Server starts with did not
have "Log on as a Service" right!

Of course, every time I re-entered the password for this
user in Services applet on either of the cluster nodes,
Windows automatically gives the account this right and so
the SQL can fail over successfully. But as soon as the
Group Policy is applied next, this right is removed, so
the next failover will not succeed.

Thanks for your assistance, Chandra.

Steve