unitproc/bktemp-count_jobs

   1 #!/usr/bin/env bash
   2
   3 # Note: GNU Parallel obviates the need for this script. The main
   4 # motivation to writing this script was to manage the number of CPU
   5 # threads being spawned by a list of files being feed to a while loop.
   6
   7 # Functions
   8 yell() { echo "$0: $*" >&2; } # print script path and all args to stderr
   9 die() { yell "$*"; exit 111; } # same as yell() but non-zero exit status
  10 try() { "$@" || die "cannot $*"; } # runs args as command, reports args if command fails
  11 count_jobs() {
  12     local job_count;
  13     job_count="$(jobs -r | wc -l | tr -d ' ' )";
  14     #yell "DEBUG:job_count:$job_count";
  15     if [[ -z $job_count ]]; then job_count="0"; fi;
  16     echo "$job_count";
  17 }; # Return number of background jobs
  18 test_job() {
  19     #yell "DEBUG:starting test_job() with:$1";
  20     # Do work on $1 file
  21     if [[ -f $1 ]]; then cat "$1" 1>/dev/random 2>&1; fi; # read file
  22     sleep "$(shuf -i1-10 -n1)"; # debug
  23 }; # Test job
  24 count_jobs_display_update() {
  25     # Depends: various variables
  26     if [[ $(( SECONDS % jobs_update_interval )) -eq $jobs_update_init_delay ]] && \
  27        [[ $permit_update == "true" ]]; then
  28     jobs_beg="$jobs_n";
  29     jobs_run="$(count_jobs)";
  30     jobs_end="$((jobs_beg - jobs_run))";
  31     yell "STATUS:$jobs_beg jobs begun. $jobs_end jobs ended. $jobs_run jobs running.";
  32     permit_update="false";
  33     fi;
  34     if [[ $(( SECONDS % jobs_update_interval )) -eq $((jobs_update_init_delay + 1)) ]]; then
  35         permit_update="true";
  36     fi;
  37 }; # periodically display updates
  38
  39
  40 # Test Code
  41 path_target="/tmp"; # path to dir with files to run test_job() on
  42
  43 ## Setup file list
  44 list_paths="$(find "$path_target" -type f 2>/dev/null)";
  45
  46 ## Perform test_job() on each file
  47 jobs_max=100; # adjust me (e.g. "4" on 4-core CPU)
  48 jobs_update_interval=5; # seconds.
  49 jobs_update_init_delay=1; # seconds.
  50 permit_update="true"; # flag to make updates happen periodically
  51 jobs_n=0; # loop counter
  52 jobs_check_delay=0.001; # seconds between each job count check if running jobs > jobs_max
  53 while read -r line; do
  54     ((jobs_n++));
  55     #yell "DEBUG:jobs_n:$jobs_n"
  56
  57     ## Wait until job count falls below $jobs_max
  58     while [[ "$(count_jobs)" -gt $jobs_max ]]; do
  59         #yell "DEBUG:sleeping since $(count_jobs) > $jobs_max";
  60         sleep "$jobs_check_delay";
  61         count_jobs_display_update;
  62     done;
  63
  64     ## Start new parallel job on file $line
  65     test_job "$line" 1>/dev/null 2>&1 &
  66
  67     ## Get job status updates every $jobs_update_interval seconds
  68     count_jobs_display_update;
  69
  70     #sleep 1; # debug
  71 done < <( shuf < <(echo -n "$list_paths") );
  72 yell "STATUS:All jobs ($jobs_n) started. $(count_jobs) jobs running.";
  73
  74 # Detect when no outstanding jobs
  75 while [[ "$(count_jobs)" -gt 0 ]]; do
  76     sleep 1;
  77     count_jobs_display_update;
  78 done;
  79 yell "STATUS:No more jobs visible.";
  80
  81 # Author: Steven Baltakatei Sandoval
  82 # License: GPLv3+