HFSS15: Troubleshooting for LSF

Ensure that the LSF ’lsun’ command is enabled:

Make sure there is no user error. For example:

• Use correct path to HFSS and full path to project

• Ensure sufficient resources (cpu/memory/disk) are allocated to job: check space in temp directory, space in project directory (consider quotas)

• Ansoft project to be available on the execution host

• Ensure job submitter has permissions to read/write to project directory and read/execute permissions to installation directory

• Check for locked project

Find if this is a standalone product issue:

Run HFSS interactively on the machine, outside of the scheduler, and see if it comes up and analyzes.

Send the following logs to development:

• Output of LSF batch job corresponding to the analysis of Ansoft project: You can obtain this using LSF commands: use bacct if job’s output is not redirected to a file. For example,

“bacct –l <jobid>”

• Batch log generated by Ansoft product (typically, projectname.log, in the same directory as the Ansoft project file)

• Debug logs generated by Ansoft product, during the running of LSF job.

Set ANSOFT_DEBUG_MODE to 1

It is good to collect another set of logs with above value set to 7

Create \shared_driveansdebug directory that is accessible by all machines in the cluster

Set ANSOFT_DEBUG_LOG to a file in \shared_driveansdebug’ directory, for example,

\shared_driveansdebuganslog

Set ANSOFT_DEBUG_LOG_SEPARATE to 1.

Set ANSOFT_LSF_LOG to a distinct file in \shared_driveansdebug’ directory, for example,

\shared_driveansdebuglsf.log

• For each pair of the machines, between which the remote analysis fails, run "ping remote-machine" and note the output

• For each machine in the network, dump the network interfaces (for example, run "ifconfig -a") and note the output

• Email all 5 logs to development

- Job’s log generated by LSF

- Log generated by Ansoft product

- Entire \shared_driveansdebug folder

- Output from ping command

- Output related to compute node’s network interfaces