From 5d49965bb99756cd815a562960c1cc1ae5d396a4 Mon Sep 17 00:00:00 2001 From: Michal Berger Date: Tue, 28 Dec 2021 15:52:28 +0100 Subject: [PATCH] test/scheduler: Ignore failures while removing dummy cgroups Upon removal, there's a small chance that new threads would jump into these cgroups AFTER we read the list of PIDs to migrate. In such a case, there may be some processes left preventing us from finally removing the cgroup. Since we can't fully control that without continuously reading the list, simply ignore it. Fixes issue #2298 Signed-off-by: Michal Berger Change-Id: I0d50b975742a8b15edd25c6b4e51ab337b017eaa Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/10872 Community-CI: Broadcom CI Reviewed-by: Tomasz Zawadzki Reviewed-by: Shuhei Matsumoto Reviewed-by: Konrad Sztyber Tested-by: SPDK CI Jenkins --- test/scheduler/isolate_cores.sh | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/test/scheduler/isolate_cores.sh b/test/scheduler/isolate_cores.sh index 5b272c1e7..d32cac0fb 100644 --- a/test/scheduler/isolate_cores.sh +++ b/test/scheduler/isolate_cores.sh @@ -7,8 +7,8 @@ restore_cgroups() { xtrace_disable kill_in_cgroup "/cpuset/spdk" remove_cgroup "/cpuset/spdk" - remove_cgroup "/cpuset/all" - remove_cpuset_cgroup + remove_cgroup "/cpuset/all" || true + remove_cpuset_cgroup || true xtrace_restore }