Hello,
it looks to me like warewulf is creating nodes (when having populated the "eth0.FQDN" field in the node definition via "--fqdn=") that resolve "hostname" lookups to their FQDN, instead of the expected "hostname -s".
This causes problems for NHC when triggered by SLURM:
error: _find_node_record(763): lookup failure for <FQDN>
Is there any option i am not aware of to come up with WareWulf provisioning nodes which resolve "hostname" lookups to "hostname -s" equivalents?
It looks like warewulf3 graps the hostname from the first position in /etc/host, which is the FQDN, when dropping "localdomain" as described.
Extracting the short hostname would resolve the problem.
Currently i have to readjust the hostnames of the deployed compute nodes using a postinstall script, which is not as neat as it could be.
Best
Hello,
it looks to me like warewulf is creating nodes (when having populated the "eth0.FQDN" field in the node definition via "--fqdn=") that resolve "hostname" lookups to their FQDN, instead of the expected "hostname -s".
This causes problems for NHC when triggered by SLURM:
error: _find_node_record(763): lookup failure for <FQDN>Is there any option i am not aware of to come up with WareWulf provisioning nodes which resolve "hostname" lookups to "hostname -s" equivalents?
It looks like warewulf3 graps the hostname from the first position in /etc/host, which is the FQDN, when dropping "localdomain" as described.
Extracting the short hostname would resolve the problem.
Currently i have to readjust the hostnames of the deployed compute nodes using a postinstall script, which is not as neat as it could be.
Best