Merge pull request #1038 from trws/fix-segfault

planner: ensure result in planner_avail_resources_at

resource/planner/planner.cpp

@@ -519,7 +519,7 @@ extern "C" int64_t planner_avail_resources_during (planner_t *ctx,
 extern "C" int64_t planner_avail_resources_at (planner_t *ctx, int64_t at)
 {
     scheduled_point_t *state = nullptr;
-    if (!ctx || at > ctx->plan_end) {
+    if (!ctx || at > ctx->plan_end || at < ctx->plan_start) {
         errno = EINVAL;
         return -1;
     }

t/Makefile.am

@@ -97,6 +97,7 @@ TESTS = \
     t4010-match-conf.t \
     t4011-match-duration.t \
     t5000-valgrind.t \
+    t5100-issues-test-driver.t \
     t6000-graph-size.t \
     t6001-match-formats.t \
     t6002-graph-hwloc.t \
@@ -105,6 +106,9 @@ TESTS = \
 
 check_SCRIPTS = $(TESTS)
 
+dist_check_SCRIPTS = \
+	issues/t1035-fluxion-reload.sh
+
 EXTRA_DIST= \
 	$(check_SCRIPTS) \
 	data \

t/issues/t1035-fluxion-reload.sh

@@ -0,0 +1,69 @@
+#!/bin/bash
+#
+#  Ensure fluxion modules can recover running jobs with rv1 match format.
+#
+log() { printf "issue#1035: $@\n" >&2; }
+die() { log "$@"; exit 1; }
+run_timeout() {
+    "${PYTHON:-python3}" "${SHARNESS_TEST_SRCDIR}/scripts/run_timeout.py" "$@"
+}
+
+if test -z "$ISSUE_1035_TEST_ACTIVE"; then
+    export ISSUE_1035_TEST_ACTIVE=t
+    log "relaunching under test instance of size 4..."
+    exec flux start -s 4 $0 "$@"
+fi
+test $(flux resource list -no {nnodes}) -eq 4 || die "test requires 4 nodes"
+
+log "Unloading modules..."
+flux module remove sched-fluxion-qmanager
+flux module remove sched-fluxion-resource
+flux module remove resource
+
+log "Amending instance resource set with properties: batch, debug..."
+flux kvs get resource.R \
+	| flux R set-property batch:0-1 debug:2-3 \
+	| flux kvs put -r resource.R=-
+#flux kvs get resource.R | jq
+
+log "Loading config with queues and match-format=\"rv1\"..."
+flux config load <<EOF
+[queues.debug]
+requires = ["debug"]
+
+[queues.batch]
+requires = ["batch"]
+
+[sched-fluxion-resource]
+match-format = "rv1"
+EOF
+flux config get | \
+	jq -e ".\"sched-fluxion-resource\".\"match-format\" == \"rv1\"" \
+	|| die "failed to set sched-fluxion-resource.match-format = rv1"
+
+log "Reloading modules..."
+flux module load resource noverify
+flux module load sched-fluxion-resource
+flux module load sched-fluxion-qmanager
+flux dmesg -HL | grep version | tail -2
+
+log "Starting all queues..."
+flux queue start --all --quiet
+flux queue status
+flux resource list -s free
+
+log "Submitting two sleep jobs..."
+run_timeout 10 flux submit -N2 --wait-event=start --queue=debug sleep inf
+run_timeout 10 flux submit -N2 --wait-event=start --queue=batch sleep inf
+
+log "Reloading fluxion..."
+flux module unload sched-fluxion-qmanager
+flux module reload sched-fluxion-resource
+flux module load sched-fluxion-qmanager
+
+log "Checking that running jobs were recovered..."
+flux jobs -ano "{id.f58:>12} {status_abbrev:>2} {name}"
+test $(flux jobs -no {id} | wc -l) -eq 2 \
+	|| die "Expected 2 jobs still running"
+flux cancel --all
+

t/t5100-issues-test-driver.t

@@ -0,0 +1,31 @@
+#!/bin/sh
+#
+test_description='Verify that fixed issues remain fixed'
+
+. `dirname $0`/sharness.sh
+
+if test_have_prereq ASAN; then
+    skip_all='skipping issues tests under AddressSanitizer'
+    test_done
+fi
+skip_all_unless_have jq
+
+SIZE=2
+test_under_flux ${SIZE}
+echo "# $0: flux session size will be ${SIZE}"
+
+if test -z "$T5100_ISSUES_GLOB"; then
+    T5100_ISSUES_GLOB="*"
+fi
+
+flux bulksubmit -n1 -o pty --job-name={./%} -t 10m \
+	--flags=waitable \
+	--quiet --watch  \
+	flux start {} \
+	::: ${SHARNESS_TEST_SRCDIR}/issues/${T5100_ISSUES_GLOB}
+
+for id in $(flux jobs -ano {id}); do
+    test_expect_success $(flux jobs -no {name} $id) "flux job attach $id"
+done
+
+test_done