Moab Workload Manager
1. "ERROR:user [user](UID:500) is not authorized to run this program"
2. Not all nodes are showing up when I use showq or mdiag -n
3. Moab shows 0 of 0 nodes on my system
4. "ERROR: cannot migrate job to PBS - exec cannot be located or is not an executable"
5. "ERROR: cannot migrate job to PBS - user 'root' does not exist on local system"
6. How to set email notification
7. How do I uninstall Moab Workload Manager evaluation to install the production version?
General Troubleshooting Tips
View the Flash demonstration for diagnosing Moab
1. Problem:
"ERROR:user [user](UID:500) is not authorized to run this program“
When running Moab I receive the error, "ERROR: user [user](UID:500) is not authorized to run this program."
Solution:
1. The user needs to have ADMIN1 rights.
Open moab.cfg for editing.
Add the user as the primary admin next to USERS= with a comma separating other users.
ADMINCFG[1] USERS=greg,thomas |
2. Problem:
Not all my nodes are showing up under showq or mdiag -n.
Solution:
1. Check to see if your resource manager displays the nodes.
With TORQUE running, type pbsnodes -a.
If all the nodes do not display, see Question 1 of the TORQUE common issues.
2. Check your license file, moab.lic, and make sure that it is created for the correct amount of processors on your cluster. Moab will only display the processors allotted in the license.
If you need Moab allocated for more processors, please contact Cluster Resources.
3. Problem:
Moab shows 0 of 0 nodes on my system
When I run showq it states I have 0 of 0 processors on 0 of 0 nodes, but my resource manager displays all my nodes. What is the issue?
Solution:
1. If you have this problem, when you run mdiag -R you will see a similar output showing the state of the resource manager as down.
RM[base] State: Down |
2. If TORQUE was installed into a non-default location and/or the TORQUE server files, pbs_server and pbs_iff, are not configured in your environment PATH, then Moab will not know where to locate TORQUE.
In moab.cfg, add the parameter RMCFG[RM name] SBINDIR=[location of torque server (ie. pbs_Server and pbs_iff)].
• Restart Moab and check to see if showq displays nodes.
| RMCFG[base] SBINDIR=/usr/local/torque/sbin |
3. If you machine has hyperthreading turned on, Moab needs to be built on a machine with hyperthreading turned on or the resource manager will not communicate properly with Moab.
If this is the case in your configuration, please contact Cluster Resources for a correct build.
4. Problem:
"ERROR: cannot migrate job to PBS - exec cannot be located or is not an executable"
When I submit a job using msub through Moab Workload Manager or submitting through Cluster Manager or Access Portal I receive the error: "ERROR: cannot migrate job to PBS - exec cannot be located or is not an executable."
Solution:
1.Make sure that the location of qsub is in your environment PATH for Moab to locate.
2. For msub to properly work, you must run Moab as root.
3. If Moab still cannot locate qsub, add the following parameter to moab.cfg:
- RMCFG[RM name] SUBMITCMD=[location of qsub]/qsub
- Restart Moab and try submitting a job using msub.
| RMCFG[base] SUBMITCMD=/usr/local/bin/qsub |
5. Problem:
"ERROR: cannot migrate job to PBS - user 'root' does not exist on local system"
When I submit a job using msub through Moab Workload Manager I receive the error: "ERROR: cannot submit job as root."
Solution:
1. Moab does not allow job submissions from root. Switch to another user and submit the job again.
6. Problem:
How/Where do I set the email notification portion of Moab?
Solution:
1. Moab can e-mail people about events it creates (starting Moab, shutting down Moab, etc.) but e-mail about job events is handled at the resource manager level. For example, if using TORQUE, documentation on e-mail settings can be found here
You will have to reference your resource manager documentation if not using TORQUE/PBS.
7. Problem:
How do I uninstall Moab Workload Manager evaluation to install the production version?
Solution:
1. The eval version of Moab is installed using :make install" and uninstalled using "make uninstall". It should be noted that if you are using the default locations: binaries installed to usr/local{bin|sbin}, Moab home directory in /opt/moab, you are safe to simply install the newer version and it will over write the eval versions of the binaries while preserving your moab.cfg.


