I am trying to link two computers (both centos 7) using PBS for running and scheduling mpi jobs. I am using first computer as head node and also as execution node and client node. I have all the services (pbs_comm, pbs_mom, pbs_sched, pbs_server.bin) working and I can submit jobs to default queue. I have installed pbs_execution on the second machine and verified pbs_mom is running. Using pbs_hostn I verified i can see server. But when I do "pbsnodes -a " i keep getting error:
No route to host
pbsnodes: cannot connect to server manowar, error=113
Can someone tell me how to fix this?
thanks,
Shashi
Please find this checklist on system configuration on your two computers
SELinux disabled and rebooted
Firewalls disabled
Both machines have static IP address/hostnames defined and mentioned in /etc/hosts
The contents of /etc/hosts are correct on both the systems.
Passsword-less SSH is configured between those two systems and to themselves( hostbased passwordless authentication or ssh keybased authentications) for all the users
Common home directory via NFS or Common Working directory which is readable writable and accessible by all the users across both the systems.
This might be related to
Network resolutions issue , check /etc/hosts
Check the pbs services are running systemctl status pbs