On Mar 12, 2008, at 2:31 PM, Greg Thain wrote:
A couple of folks have asked about SCHEDD_NAME in light of schedd failover. I (and Todd) thought that they way it was supposed to workwas that you'd set SCHEDD_NAME explicitly in both the active and backupschedd to the same value. However, it looks this doesn't work, the schedd always replaces the string after the "@" with the fqdn. Is there any wisdom on why this is?
All I can remember is that back in the dark ages, when we were first messing around with "personal condors" and trying to make things work- able, we found that people were regularly confused and ran into trouble if everything after the @ wasn't a fqdn.
Furthermore, there is logic in The Tool(tm) that resolves hostnames in the command-line args to be able to locate the daemon you're trying to talk to. Actually, the logic is probably in the Daemon object itself. So, if you say:
"condor_reschedule wright@tonic"the tool knows enough to find the right schedd. That depends on it really being a fqdn...
I won't claim these are Good Reasons(tm), but that's what I remember about why it's like this.
So, any change to the schedd (or master, startd, etc) that changes how the FOO_NAME is set could potentially have implications for the command-line tools or other code using Daemon objects.
Cheers, -Derek