Testing the Batchsystem Server
Since no worker nodes are available to the batch system server until now, only a limited set of functions can be tested at this very moment. At least you can check if the yaim configuration step was successful.
- Check if the hostname of the batch system server is set properly in the file /var/spool/pbs/server_name
- Check if the list of worker nodes is set in the file /var/spool/pbs/server_priv/nodes
- Check if the Torque process is running
ps aux | grep -i pbs
- Check if the Maui process is running
ps aux | grep -i maui
In the case that one of the processes is not running, please start it
/etc/init.d/maui start /etc/init.d/pbs_server
If the pbs process is running you can execute qmgr on the command line. This command allows you to manage queues manually. To display the configuration of the batch system server enter
and to list only the configuration of a single queue
list queue <queue_name>
The tools described next can only be tested if the worker node has already been installed and the torque and maui processes are running.
Check the status of the available worker nodes by issuing the pbsnodes command
pbsnodes -l #shows the offline nodes
As output you should receive that all worker nodes configured in your worker node list are down:
Print the attributes for a single worker node
pbsnodes -a <hostname of the worker node>
Print queue information via the qstat command (i. e. for a queue dech)
qstat -Q test Queue Max Tot Ena Str Que Run Hld Wat Trn Ext T ---------------- --- --- --- --- --- --- --- --- --- --- - test 100 0 yes yes 0 0 0 0 0 0 E ----
Try to figure what the following commands do:
Remember some of them! You can uses them later when submitting jobs from the CREAM compute element to the batchsystem.
Now you can proceed to the batch system server administration