Should the Prologslurmctld script only run after the Epilogslurmctld script finishes?
Below you can see JobA runs and completes. While Epilogslurmctld (from JobA Node A) is executing on the Slurm controller the Prologslurmctld script for the next job (from Job B Node A) is also running on the Slurm controller. This breaks our workflow as we are expecting the next Prologslurmctld script to only run when the prior job is 100% completed (initial Prologslurmctld completes AND Job completes AND Job Epilogslurmctld completes). Prologslurmctld > Here Prolog is starting for Job B (ID 812) *2021-09-27 15:42:58,746* | INFO | Starting... Epilogslurmctld > Here Epilog is starting and ending for Job A (ID 811) *2021-09-27 15:42:56,694* | INFO | Starting... *2021-09-27 15:43:01,756* | INFO | Exiting 0 after main [2021-09-27T15:42:50.224] debug: sched/backfill: _attempt_backfill: beginning [2021-09-27T15:42:50.224] debug: sched/backfill: _attempt_backfill: 1 jobs to backfill [2021-09-27T15:42:56.653] _job_complete: JobId=811 WEXITSTATUS 0 [2021-09-27T15:42:56.653] debug: email msg to root: Slurm Job_id=811 Name=JobA_BIOS_fixedfreq_1067mclk_nps1.ini Ended, Run time 00:22:36, COMPLETED, ExitCode 0 *[2021-09-27T15:42:56.657]* _job_complete: JobId=811 done [2021-09-27T15:42:58.703] debug: sched: Running job scheduler for full queue. *[2021-09-27T15:42:58.704]* debug: email msg to root: Slurm Job_id=812 Name=JobA_BIOS_fixedfreq_1600mclk_nps1.ini Began, Queued time 00:22:38 [2021-09-27T15:42:58.704] sched: Allocate JobId=812 NodeList=xxx #CPUs=256 Partition=xxx
