Error monitoring Clustered SQL server

tdietrich

Getting the error: Unreachable (Cannot connect)

The log reads as follows:
18 Jun 2012 9:12 AM SQL Ping Connected
18 Jun 2012 9:12 AM Registry OpenRegistryHive: PerformanceData Cannot connect Win32Exception The network path was not found
18 Jun 2012 9:12 AM Registry OpenRegistryHive: LocalMachine Cannot connect Win32Exception The network path was not found

What is the typical cause of this issue? We've got two two box clusters configured in similar fashions, and one works while the other shows this error.

tdietrich

I've ran though every step in this test:

http://www.red-gate.com/supportcenter/C ... 469857.htm

They all worked properly. Now I'm really stumped.

Edit:

Actually, I can not remote connect to the registry at the SQL cluster's address, but I can at the Windows cluster, the DTC cluster, or either node of the cluster's address.

Brian Donahue

SQL Monitor wants to get data from the cluster node rather than the IP address configured as the cluster resource.

When you add a cluster to SQL Monitor, it's importand to specify the server using the node name. It works best to just add all instances. Say you have a cluster with two nodes, NODEA and NODEB. You would add NODEA\*. SQL Monitor will query the registry of NODEA to work out all of the other participating nodes and configure the remote registry, SQL, and WMI connections accordingly.

tdietrich

The issue was isolated to a configuration setting preventing connections to the remote registry at the SQL cluster's address.

Our network administrator found the solution: http://social.technet.microsoft.com/Forums/eu/configmgrsetup/thread/8b4a2246-1cff-4d31-9513-c1239c1e694b

Of the tests we conducted above the root cause of the short file name error during the Pre-req checks was found to be the fact that we were not able to remotely access the registry using the SQL Virtual name which we should have been able to do. On contacting Microsoft Support we were informed that there are known issues around NoRemap registry key needing to be set if you are using Network Name resource that is dependent on a separate new IP Address resource – see http://support.microsoft.com/?id=306985

The final solution was to add a new REG_MULTI_SZ registry key to BOTH cluster nodes called:-
NoRemapPipes to:-
HKLM\SYSTEM\CurrentControlSet\Services\lanmanserver\parameters\

This key should contain the values of:-
WINREG
EVENTLOG
svcctl

After adding the above key and values we needed to reboot both cluster nodes and then we were able to connect to the SQL Virtual names (CLUSQL02 + CLUSQL01) using Regedit (Remote registry). Running the SCCM Pre –req checker now returns no errors.

Hope this soluton helps someone else.