Trouble setting up monitoring on a cluster

scubamandan

Hi. I have an active/passive cluster that I can't seem to set up. I get the following error: "Unreachable Cannot connect"

And in the logs:

GROUP EVENT OUTCOME EXCEPTION EXCEPTION MESSAGE
Registry OpenRegistryHive: LocalMachine Cannot connect Win32Exception The network path was not found

I've tried following the steps here:
https://documentation.red-gate.com/display/SM5/Testing+data+collection+methods

From the base machine (doubles as web host) I can ping the nodes as well as the Cluster name, I can also connect to the registry remotely as well as perfmon using the required user.. I'm assuming the problem is network one, but if I'm passing the tests, how I could I be failing to connect?

What else can I check? Anyone have any ideas? It'd much appreciated.

Alex B

Hi scubamandan,

Usually when there is an issue like this with a cluster this ends up being the answer - see the quote box at the bottom.

Please let us know if that helps resolve the issue!

Kind regards,
Alex

scubamandan

Thanks Alex. Though I've just checked and those keys values are already in the registry anyway.

Should I be testing from the SQL monitor service to the nodes, or to the Cluster Name? i.e. testing remote registry and WMI.

Alex B

Hi scubamandan,

You should be testing from the SQL Monitor machine as the user SQL Monitor uses to whichever element is having the problem - so if it's a node, standalone server or the cluster name itself - test to that entity.

Kind regards,
Alex

scubamandan

Thanks, but yes that's where I'm testing from. I meant where should I be testing to. The cluster name or the node host name. Though I have tested to both.

Alex B

Hi scubamandan,

Sorry, my last message was a bit unclear - I had said "...to whichever element is having the problem - so if it's a node, standalone server or the cluster name itself - test to that entity." meaning it should be to the cluster name if that was where the error was displaying - but you indicated you have tried to both the node and cluster names and it is working properly. I have just received your support ticket and will continue troubleshooting there and then post whatever resolution we find here.

Kind regards,
Alex

scubamandan

Thanks to Alex for working through this with me. We found the cause of the issue. We have more than one domain here, and I'm no network expert (far from it), but issue was caused by missing DNS entries. Our cluster name and virtual server name were in DNS against domain 1. But the monitor and the actual servers themselves to be monitored were in domain 2. When running through the checks, e.g. check remote registry and WMI, everything worked... which was confusing. This is because we didn't fully qualify our connections with the full domain when testing. i.e. myserver.2.local. As SQL Monitor was specifying the full name, it could not see the servers. Our network guys added these two extra DNS entries and it's all working now!