I am looking for help on configuring a HTCondorCE on an OpenStack instance. We are planning to setup a WLCG CE in a HPC cIuster running OpenStack. We plan to use HTCondorCE + Slurm. Currently I am using a local OpenStack instance to
understand the configuration, before I can move to the real system.
In the OpenStack the VM is using an internal address (192.168.98.137), which is translated to a so-called floating IP (192.168.120.27). This address is registered in our local DNS (liko-ce-dev.local.hephy.at). Also reverse loookup works. But evidentlyÂ internally there is a different IP as externally.
Starting condor-ce I find now error messages in various log files. Trying to do the obvious, setting
NETWORK_INTERFACE = 192.168.98.137
to my internal address does not solve the problem.
In particular I find failures of the GSI handshake, suggesting to set various config options.
But I think its obvious that there is need to configure condor for such a situation.
Did anybody experience a similar problem or are there somewhere instruction what to do in such a case.