system tests: fix race in stop test

In the unlock/timeout test, on slow systems, 'podman ps' could
catch the container before the just-backgrounded 'podman stop'
sends the signal. Wait for signal ack from container before
we inspect it.

Also: If I understand the test correctly, it wasn't actually
checking that 'ps' could grab the lock while the container
was exiting. Add a check.

Signed-off-by: Ed Santiago <santiago@redhat.com>

1 files changed, 29 insertions, 2 deletions

diff --git a/test/system/050-stop.bats b/test/system/050-stop.bats
index 2ed791429..d809507a5 100644
--- a/test/system/050-stop.bats
+++ b/test/system/050-stop.bats
@@ -119,11 +119,31 @@ load helpers
     # the container's status.
 
     run_podman run --name stopme -d $IMAGE sh -c \
-        "trap 'echo Received SIGTERM, ignoring' SIGTERM; echo READY; while :; do sleep 1; done"
+        "trap 'echo Received SIGTERM, ignoring' SIGTERM; echo READY; while :; do sleep 0.2; done"
 
-    # Stop the container in the background
+    wait_for_ready stopme
+
+    local t0=$SECONDS
+    # Stop the container, but do so in the background so we can inspect
+    # the container status while it's stopping. Use $PODMAN because we
+    # don't want the overhead and error checks of run_podman.
     $PODMAN stop -t 20 stopme &
 
+    # Wait for container to acknowledge the signal. We can't use wait_for_output
+    # because that aborts if .State.Running != true
+    local timeout=5
+    while [[ $timeout -gt 0 ]]; do
+        run_podman logs stopme
+        if [[ "$output" =~ "Received SIGTERM, ignoring" ]]; then
+            break
+        fi
+        timeout=$((timeout - 1))
+        if [[ $timeout -eq 0 ]]; then
+            die "Timed out waiting for container to receive SIGERM"
+        fi
+        sleep 0.5
+    done
+
     # Other commands can acquire the lock
     run_podman ps -a
 
@@ -131,6 +151,13 @@ load helpers
     run_podman inspect --format '{{.State.Status}}' stopme
     is "$output" "stopping" "Status of container should be 'stopping'"
 
+    # Time check: make sure we were able to run 'ps' before the container
+    # exited. If this takes too long, it means ps had to wait for lock.
+    local delta_t=$(( $SECONDS - t0 ))
+    if [[ $delta_t -gt 5 ]]; then
+        die "Operations took too long ($delta_t seconds)"
+    fi
+
     run_podman kill stopme
     run_podman wait stopme