Changeset 3869 for trunk/LMDZ.COMMON/libf/misc

Timestamp:

Jul 28, 2025, 4:56:58 PM (4 weeks ago)

Author:

jbclement

Message:

PEM:
Bug correction to detect the job time limit with PBS/TORQUE. Making it more robust and automatic for the launching script and the Fortran code.
JBC

Location:

trunk/LMDZ.COMMON/libf/misc

Files:

• job_id_mod.F90 (added)
• job_timelimit_mod.F90 (modified) (6 diffs)
• parse_args_mod.F90 (modified) (4 diffs)

trunk/LMDZ.COMMON/libf/misc/job_timelimit_mod.F90

@@ -4 +4 @@
 ! DESCRIPTION:
 !    Retrieves the time limit (in seconds) for a given job ID via
-!    SLURM (squeue) or PBS/TORQUE (qstat).
+!    SLURM or PBS/TORQUE.
 !***********************************************************************
 
@@ -17 +17 @@
 !=======================================================================
 
-SUBROUTINE get_job_timelimit(arg)
+SUBROUTINE get_job_timelimit(jobid)
 
 implicit none
 
 !---- Arguments
-character(*), intent(in) :: arg
+character(*), intent(in) :: jobid
 
 !---- Variables
@@ -33 +33 @@
 ! Check that the job ID is numeric
 num_str = .true.
-do i = 1,len_trim(arg)
-    if (arg(i:i) < '0' .or. arg(i:i) > '9') then
+do i = 1,len_trim(jobid)
+    if (jobid(i:i) < '0' .or. jobid(i:i) > '9') then
         num_str = .false.
         exit
@@ -42 +42 @@
 
 ! Try SLURM (squeue)
-call execute_command_line('squeue -j '//trim(adjustl(arg))//' -h --Format TimeLimit > tmp_cmdout.txt',cmdstat = cstat)
+call execute_command_line('squeue -j '//trim(adjustl(jobid))//' -h --Format TimeLimit > tmp_timelimit.txt',cmdstat = cstat)
 if (cstat /= 0) then
     ! On failure, try PBS/TORQUE (qstat)
-    call execute_command_line('qstat -f '//trim(adjustl(arg))//' | grep "Walltime" | awk ''{print $3}'' > tmp_cmdout.txt',cmdstat = cstat)
+    call execute_command_line('qstat -f '//trim(adjustl(jobid))//' | grep "Walltime" | awk ''{print $3}'' > tmp_timelimit.txt',cmdstat = cstat)
     if (cstat > 0) then
         error stop 'Error: command execution failed!'
@@ -54 +54 @@
 
 ! Read the output
-open(1,file = 'tmp_cmdout.txt',status = 'old')
+open(1,file = 'tmp_timelimit.txt',status = 'old')
 read(1,'(a)') chtimelimit
 close(1)
@@ -60 +60 @@
 
 ! Remove temporary file
-call execute_command_line('rm tmp_cmdout.txt',cmdstat = cstat)
+call execute_command_line('rm tmp_timelimit.txt',cmdstat = cstat)
 if (cstat > 0) then
     error stop 'Error: command execution failed!'

trunk/LMDZ.COMMON/libf/misc/parse_args_mod.F90

@@ -12 +12 @@
 
 use pgrm_version_mod,  only: print_pgrm_version
+use job_id_mod,        only: get_job_id
 use job_timelimit_mod, only: get_job_timelimit
 
@@ -30 +31 @@
 !---- Variables
 integer        :: narg, i, eq_pos
-character(256) :: arg, key, vlu
+character(256) :: arg, key, vlu, jobid
 
 !---- Code
@@ -73 +74 @@
             write(*,*) 'SSO fields will be included in "start_archive.nc"'
 
-        case ('--jobid')
-            call get_job_timelimit(vlu)
+        case ('--auto-exit')
+            call get_job_id(jobid)
+            call get_job_timelimit(jobid)
 
         case default
@@ -93 +95 @@
     write(*,*) '  --version [file]    Print program version and exit (optional output file)'
     write(*,*) '  --add-sso           Add SSO fields to "start_archive.nc" (only available for Mars start2archive)'
-    write(*,*) '  --jobid <id>        Query the time limit for the specified job ID,'
-    write(*,*) '                      enabling self-termination before timeout (only available for the PEM)'
+    write(*,*) '  --auto-exit         Enable automatic termination before reaching the job time limit (only available for the PEM)'
     write(*,*)
 END SUBROUTINE print_usage