March 10, 2009 at 6:38 am
Hey all,
- Uninstalled SQL 2000, got a message about "unable to uninstall, you'll have to manually uninstall"
- 20 minutes elapsed, then got a message "Uninstall complete, reboot now"
- Rebooted
- Check both nodes, no SQL
- SQL2005 install errored out on the virtual network name already exists
- Removed all SQL entries in the cluster SQL group
- Now the install is erroring at the very last step "the path was not found" -- no more detail
- Support.log shows this:
MSI (s) (68:9C) [08:02:10:812]: Invoking remote custom action. DLL: C:\WINDOWS\Installer\MSIB4.tmp, Entrypoint: Sqlmsirc_BackupAppGuid
Action start 08:02:10: Sqlmsirc_BackupAppGuid.0D10D66A_FB58_4290_B8AC_58FF44A6C1F1.
Failed to query registry value (SOFTWARE\Microsoft\Microsoft SQL Server\90\Bootstrap\MSIRefCount\Uninstall)
Error: The system cannot find the file specified.
-- I'm assuming this is the error it's blowing up on? Yet the log goes on for quite a bit more.
- Followed KB article to ensure SQL2000 was fully removed: http://support.microsoft.com/kb/290991#4
- Registry entries were present for one of the two nodes, removed and restarted
- Still throwing the error
- Cluster Administrator errors out stating "An error occured attempting to read properties for the 'SQL Server Agent' resource type. The specified node does not support a resource of this type. This may be due to version inconsistencies or due to the absence of the resource DLL on this node."
- Looking at resource types, the SQL Server Agent has a little blue icon next to it
- SQAGTRES.dll is not present in C:\windows\system32
- I assumed the uninstall would have removed these resources, but possibly still there because of the bad 2000 uninstall?
Not sure if the problem is in regards to the cluster admin errors or not.
Any thoughts?
March 10, 2009 at 6:52 am
Actually the issue may be because the private network in cluster admin is showing up red and disconnected ... Not sure if that was like that before or after this install attempt. Yet the initial install splash screens showed it was able to talk to the other node just fine.
We're trying to get that fixed now.
March 10, 2009 at 8:20 am
Would like to get updates as you proceed.
MJ
March 10, 2009 at 8:25 am
We fixed the private network and removed SQL Server and the Agent from the resource types ... still same problem.
March 10, 2009 at 8:58 am
Well, pretty much out of options for now ... I've tried reinstalling several times ... this is after uninstalling everything related to SQL (.net framework, native client, setup support files) and removing all the appropriate registry entries. Still same error every time.
Trying to call Microsoft now.
Anyone else got any ideas?
March 10, 2009 at 9:09 am
Adam Bean (3/10/2009)
Trying to call Microsoft now.
Adam - I was watching your other post yesterday about tips on this and wondered if this wouldn't get here. Such a messy deal trying to go through with something that is in existence.
What you are doing is, IMO, the best solution at this point.
No help but cheering you on! 🙂
David
@SQLTentmaker“He is no fool who gives what he cannot keep to gain that which he cannot lose” - Jim Elliot
March 10, 2009 at 9:10 am
Check under event viewer logs and installation log files at C:\Program Files\Microsoft SQL Server\90\Setup Bootstrap\LOG\Files.
MJ
March 10, 2009 at 9:11 am
Well, on hold now with M$ ... My money is on something in regards to the 2000 uninstall and more than likely will be a very easy fix.
March 10, 2009 at 10:06 am
if you're uninstalling the old SQL, why not just reinstall the OS as well?
March 10, 2009 at 10:09 am
Be nice if life were that easy wouldn't it be?
Anyways, while on hold with Microsoft, found that the task scheduler service is not starting on the passive node. Talking to M$ now and we're bringing in a windows guy ... pretty sure this is our issue.
March 11, 2009 at 6:05 am
Well, if anyone is interested ... Resolution was that the the Task Scheduler was the problem. Being that a cluster install uses task scheduler to install on the other node(s) and it wasn't working on the other node, was our issue. The vendor who runs their apps on this server used cluster services and changed the registry to allow for the task scheduler to run on the active node and has a dependency on the shared disk. Was able to tear down, install and put their junk back.
Fun fun.
Viewing 11 posts - 1 through 10 (of 10 total)
You must be logged in to reply to this topic. Login to reply