Merge pull request #5742 from grondo/issue#5734

do not drain ranks when job is canceled during prolog

Author: mergify[bot]

src/cmd/flux-perilog-run.py

@@ -175,8 +175,14 @@ async def run_per_rank(name, jobid, args):
         if proc.canceled:
             timeout_ids.set(rank)
             rc = 128 + signal.SIGTERM
-        elif rc != 0:
+        elif rc > 0 and rc <= 128:
+            #  process failed with non-zero exit code. Add this rank to
+            #  the failed set which will be drained.
             fail_ids.set(rank)
+        else:
+            #  process was signaled (returncode < 0) or shell reported it
+            #  was signaled (128+n). Do nothing in this case.
+            pass
         if rc > returncode:
             returncode = rc
 

t/t2274-manager-perilog.t

@@ -282,6 +282,20 @@ test_expect_success 'perilog: epilog can be specified without a prolog' '
 	flux job wait-event -t 15 $jobid epilog-start &&
 	flux job wait-event -t 15 $jobid epilog-finish
 '
+test_expect_success 'perilog: canceled prolog does not drain ranks' '
+	cat <<-EOF >config/perilog.toml &&
+	[job-manager.prolog]
+	command = [ "flux", "perilog-run", "prolog", "-vesleep,30" ]
+	EOF
+	flux config reload &&
+	flux jobtap load --remove=*.so perilog.so &&
+	jobid=$(flux submit hostname) &&
+	flux job wait-event -t 15 $jobid prolog-start &&
+	flux cancel $jobid &&
+	flux job wait-event -vt 15 $jobid prolog-finish &&
+	flux resource drain &&
+	test "$(drained_ranks)" = ""
+'
 #  Note: run this job before taking rank 3 offline below
 test_expect_success 'perilog: run job across all 4 ranks' '
 	jobid=$(flux submit --wait-event=clean -N4 -n4 true)