Problem creating vm about foreman_hyperv

[ananace]

That sounds like a really strange error, I don't do any type of manual IP lookup anywhere as far as I know. Though I guess an issue might occur if trying to use my naïve cluster command routing without proper DNS names.

How does your Compute Resource configuration look?

[selund]

I'm using ip address for the resource url. I'll try to use the dns name tomorrow, and see if that helps. Thanks for the advice.

Hmm, thinking about it, could it be that there is a function somewhere that takes the resource url, and splits in into (host,domain) using the first '.' as a separator, and when you use one node in the cluster it adds nodename.domain, as the address it tries to work with? Anyhow, I'll do a test using the fqdn tomorrow.

[ananace]

Hmm, thinking about it, could it be that there is a function somewhere that takes the resource url, and splits in into (host,domain) using the first '.' as a separator, and when you use one node in the cluster it adds nodename.domain, as the address it tries to work with?

Like I said, probably an issue with my naïve cluster command router. 😅
Really need to rethink that whole thing, but it was a simple way to get something that would run and not explode the moment there was another machine available on the cluster.

The issue is that Hyper-V - and Failover Cluster - only talk about computers in the scope of hostnames, so without WinRM authentication passthrough (really unsafe) or CredSSP (manual setup) I need to be able to spin up a separate connection to the second machine - of which I only know the hostname.
A more successful - naïve - solution might be to instead let the machine that you're connected to do a DNS lookup and return the IP for the other computer. Or just maybe promoting Kerberos connections more.

[selund]

Using fqdn for the winRM endpoint solved the problem for me.
Add a comment to use fqdn and close the ticket?

[ananace]

I mean, there only reason why IPs can't be used is due to a hack for doing direct connections to clustered machines, but I guess that would be better done through a new issue instead of leaving this one about.

I'll just throw in a disclaimer for now then.