|  | #!/usr/bin/env expect | 
|  | ############################################################################ | 
|  | # Purpose: Stress test with maximum slurmctld message concurrency. | 
|  | #          We start a large number of job steps, the simultaneously | 
|  | #          cancel them all. Message can get lost/retransmitted, so | 
|  | #          there is a delayed test for job step removal. | 
|  | ############################################################################ | 
|  | # Copyright (C) 2002-2006 The Regents of the University of California. | 
|  | # Produced at Lawrence Livermore National Laboratory (cf, DISCLAIMER). | 
|  | # Written by Morris Jette <jette1@llnl.gov> | 
|  | # CODE-OCEC-09-009. All rights reserved. | 
|  | # | 
|  | # This file is part of Slurm, a resource management program. | 
|  | # For details, see <https://slurm.schedmd.com/>. | 
|  | # Please also read the included file: DISCLAIMER. | 
|  | # | 
|  | # Slurm is free software; you can redistribute it and/or modify it under | 
|  | # the terms of the GNU General Public License as published by the Free | 
|  | # Software Foundation; either version 2 of the License, or (at your option) | 
|  | # any later version. | 
|  | # | 
|  | # Slurm is distributed in the hope that it will be useful, but WITHOUT ANY | 
|  | # WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS | 
|  | # FOR A PARTICULAR PURPOSE.  See the GNU General Public License for more | 
|  | # details. | 
|  | # | 
|  | # You should have received a copy of the GNU General Public License along | 
|  | # with Slurm; if not, write to the Free Software Foundation, Inc., | 
|  | # 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301  USA. | 
|  | ############################################################################ | 
|  | source ./globals | 
|  |  | 
|  | set file_in      "$test_dir/input" | 
|  | set job_cnt      10 | 
|  | set delay        10 | 
|  | set job_name     $test_name | 
|  | set sleep_time   300 | 
|  | set task_cnt     60 | 
|  | set job_list    [list] | 
|  |  | 
|  | proc cleanup {} { | 
|  | global job_list | 
|  |  | 
|  | cancel_job $job_list | 
|  | } | 
|  |  | 
|  | if {[param_contains [get_config_param "SlurmctldParameters"] "enable_stepmgr"]} { | 
|  | skip "This test can't because squeue --steps doesn't work for stepmgr jobs" | 
|  | } | 
|  |  | 
|  | if {![param_contains [get_config_param "SelectTypeParameters"] "CR_*MEMORY"]} { | 
|  | set mem_per_step 0 | 
|  | } else { | 
|  | set mem_per_step 8 | 
|  | } | 
|  |  | 
|  | # | 
|  | # A single slurmd can't handle a large task count without | 
|  | # running out of memory and pthreads | 
|  | # | 
|  | if {$enable_memory_leak_debug != 0} { | 
|  | set task_cnt 2 | 
|  | } | 
|  |  | 
|  | # | 
|  | # Build input script file | 
|  | # NOTE: The initial sleep is so that all of the submissions have time | 
|  | #   to occur before contending with a multitude of job step creations. | 
|  | #   This is especially important on very slow systems (e.g. AIX). | 
|  | # NOTE: Explicitly set a small memory limit. Without explicitly setting the step | 
|  | #   memory limit, it will use the system default (same as the job) which may | 
|  | #   prevent the level of parallelism desired. | 
|  | # | 
|  | make_bash_script $file_in " | 
|  | $bin_sleep 5 | 
|  | ulimit -u `ulimit -u -H` | 
|  | for ((inx=0; inx < $task_cnt; inx++)) ; do | 
|  | $srun -N1 -n1 --mem=0 --overlap $bin_sleep $sleep_time & | 
|  | done | 
|  | $srun -N1 -n1 --mem=0 --overlap $bin_sleep $sleep_time | 
|  | " | 
|  |  | 
|  | # | 
|  | # Initiate $job_cnt batch jobs | 
|  | # | 
|  | set timeout 30 | 
|  | for {set inx 0} {$inx < $job_cnt} {incr inx} { | 
|  | set job_id [submit_job -fail "--gres=craynetwork:0 --job-name=$job_name --mem-per-cpu=[expr $mem_per_step * 10] --output=/dev/null --error=/dev/null -t5 $file_in"] | 
|  | lappend job_list $job_id | 
|  | } | 
|  | log_debug "All $job_cnt jobs submitted" | 
|  |  | 
|  | # Wait for at least $job_cnt jobs to be started | 
|  | # Because we want an external variable set with the match count, it is | 
|  | # simpler to use wait_for here than wait_for_command | 
|  | set job_count 0 | 
|  | if [ | 
|  | wait_for -timeout $delay {$job_count >= $job_cnt} { | 
|  | set job_count [ | 
|  | regexp -all $job_name [ | 
|  | run_command_output -fail "$squeue --state R --name $job_name" | 
|  | ] | 
|  | ] | 
|  | } | 
|  | ] { | 
|  | log_warn "Not all jobs were started ($job_count < $job_cnt). This is ok as long as it is at least 1" | 
|  | } | 
|  | if {$job_count < 1} { | 
|  | fail "No jobs were started" | 
|  | } | 
|  |  | 
|  | # | 
|  | # There could be hundreds of job steps, we don't want to see | 
|  | # the details, but want to make sure that we did start many | 
|  | # | 
|  | set desired_tasks [expr $task_cnt * 2 / 3] | 
|  |  | 
|  | # We want to see a decent number of steps running | 
|  | if [ | 
|  | wait_for {$step_count >= $desired_tasks} { | 
|  | set step_count [ | 
|  | regexp -all sleep [ | 
|  | run_command_output -fail "$squeue --steps --name $job_name" | 
|  | ] | 
|  | ] | 
|  | } | 
|  | ] { | 
|  | fail "Only started $job_count jobs and $step_count steps. We expected at least $desired_tasks and possibly hundreds" | 
|  | } | 
|  |  | 
|  | log_debug "We found $job_count jobs and $step_count steps" |