FLUENT 6 - Checkpointing multiple jobs in a job specific manner
|
When there are two or more FLUENT sessions being run on the same machine by the same user, having check-fluent or exit-fluent files in /tmp or C:temp will checkpoint only one of the multiple jobs. This job will be the one which first completes the next iteration, from the instance the checkpoint or exit files were created. Thus, the checkpointed job will be selected on a somewhat random basis. Also, once this randomly selected job completes checkpointing, it will remove the checkpointing and exit files. Therefore, no other jobs can be checkpointed. The solution for this problem is recommended below: Add the following files to your journal file or .fluent file (in case you need these changes to apply to all the FLUENT sessions). ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; (define checkpoint/dir-name "/tmp/") ;; Replace the string with any desired directory name. (define this-job-checkpoint-filename (string-append "check-fluent-" (format #f "~a" (cx-client-id)))) (define this-job-exit-filename (string-append "exit-fluent-" (format #f "~a" (cx-client-id)))) (set! checkpoint/check-filename (string-append checkpoint/dir-name this-job-checkpoint-filename)) (set! checkpoint/exit-filename (string-append checkpoint/dir-name this-job-exit-filename)) (format "nCheckpoint filename set to: ~anExit filename set to: ~an" checkpoint/check-filename checkpoint/exit-filename) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; This will generate the job-id specific checkpoint and exit filenames and will report it to the user. Note that care must be taken to create the checkpoint and exit files with exactly the same name as is reported. |
||
|