Here is an interesting one.

2-node cluster, Windows 2003 EE x64, multi-instanced SQL Server 2005 SP2 +
CUP 7.

When we install SP3, all instances report successfull install; however,
binaries for 1 instance fail to install on 2nd inactive node. When applying
post-SP3 CUP1, we get an error message that for this instance, the passive
node does not meet the prerequisite to apply SP3 CUP1 (because the binaries
there never successfully upgraded to SP3).

From the SP3 Summary log on node 1 reports success. From the SP3 Hotfix log
on node 1, indicate the remote installation was loaded, started, and reports
success, below.

Node 1 Summary.txt:

----------------------------------------------------------------------------------
Product : Database Services (LIMS)
Product Version (Previous): 3239
Product Version (Final) : 4035
Status : Success
Log File : C:\Program Files\Microsoft SQL Server\90\Setup
Bootstrap\LOG\Hotfix\SQL9_Hotfix_KB955706_sqlrun_s ql.msp.log
Error Number : 0
Error Description :
----------------------------------------------------------------------------------

Node 1 Hotfix.txt:

02/22/2009 10:19:04.898 Installing product: SQL9
02/22/2009 10:19:04.898 Installing instance: LIMS
02/22/2009 10:19:04.898 Enumerating passive cluster nodes
02/22/2009 10:19:04.929 Patching available passive node: SQLPR2SGF
02/22/2009 10:19:04.929 Patching passive node: SQLPR2SGF
02/22/2009 10:19:04.929 Waiting for first completed passive node
02/22/2009 10:19:04.960 Task Scheduler: Created remote folder for product
instance target \\SQLPR2SGF
02/22/2009 10:19:16.445 Task Scheduler: Transferred payload to remote
product instance target \\SQLPR2SGF
02/22/2009 10:19:16.445 Task Scheduler: Transferred payload to remote
product instance target \\SQLPR2SGF
02/22/2009 10:19:16.460 Task Scheduler: Created new scheduled task for
product instance target \\SQLPR2SGF
02/22/2009 10:19:16.460 Task Scheduler: SaveTaskToDisk(): Got task interface
02/22/2009 10:19:23.445 Task Scheduler: Saved new scheduled task for product
instance target \\SQLPR2SGF
02/22/2009 10:19:23.445 Task Scheduler: Created scheduled task for product
instance target \\SQLPR2SGF
02/22/2009 10:19:23.445 Task Scheduler: Started scheduled task for product
instance target \\SQLPR2SGF
02/22/2009 10:19:23.445 Task Scheduler: Started scheduled task for product
instance target \\SQLPR2SGF
02/22/2009 10:19:26.461 Task Scheduler: Scheduled task for product instance
target has completed
02/22/2009 10:19:26.461 Task Scheduler: Waiting for exit code from scheduled
task for product instance target \\SQLPR2SGF
02/22/2009 10:19:31.461 Task Scheduler: Received exit code 0 from scheduled
task for product instance target \\SQLPR2SGF
02/22/2009 10:19:31.461 Task Scheduler: Result code for scheduled task for
product instance target has been received
02/22/2009 10:19:31.461 Task Scheduler: Removed scheduled task for product
instance target \\SQLPR2SGF
02/22/2009 10:19:31.679 Task Scheduler: Removed remote folder for product
instance target \\SQLPR2SGF
02/22/2009 10:19:31.679 Remote process completed for product instance target
02/22/2009 10:19:31.679 Exit code for passive node: SQLPR2SGF = 0
02/22/2009 10:19:34.929 First successful passive node has been completed

However, the Task Scheduler on node 2 reports the failure, and the Summary
and Hotfix logs are never created.

Node 2 Task Scheduler SchedLgU.Txt:

"SqlNodeInstall.job" (hotfix.exe) 2/22/2009 10:19:30 AM ** WARNING **
Invalid working directory..
The specific error is:
0x00000002: The system cannot find the file specified.
Verify that the directory exists and try again.
"SqlNodeInstall.job" (hotfix.exe) 2/22/2009 10:19:31 AM ** ERROR **
Unable to start task.
The specific error is:
0x80070005: Access is denied.
Try using the Task page Browse button to locate the application.
"Task Scheduler Service" 2/22/2009 10:19:31 AM ** Error **
An error has occurred that will negatively affect the operation of the
service.
The specific error is:
0x80070002: The system cannot find the file specified.
Try using the Task page Browse button to locate the application.

How come the job reports failure, but the Hotfix log reports success? How
come the Task Schudler is unable to remotely launch the upgrade for this
instance, but completes successfully for the 1st and remaining 3 instances?

Sincerely,


Anthony Thomas

That is a broken install. You can try and fail over to the broken node and
complete the install from there. Most likely you are in a catch-22 where
you need to apply SP3 but you can't because of CU1 and vice-versa.

You may have to resort to this procedure to fix the whole thing:
http://weblogs.sqlteam.com/geoffh/ar...Breakable.aspx

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



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

Thanks Geoff, but I wasn't looking for a solution. I just have to reapply
the patch to the failed node.

What bothers me is not that it failed, but that it reported success to the
active node, which continued on with the installation as if nothing incorrect
had happened.

There were no messages anywhere in the logs or Event Logs that would have
indicated failure. You had to specifically go to the remote node and
purposely look up the Task Scheduler logs to find the issue. Moreover, it
wasn't until applying the subsequent CUP that any indication was given that a
failure had actually occurred.

This is a bug in the remote installation process and the MSI package.


"Geoff N. Hiten" wrote:

Definitely a pro0blem with the installation. The cluster installer has bee
a pain for the last few versions. I like the new 2008 approach. Mich more
stable.


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


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