autotest: Attempt to cleanup SPDK processes prior running tests

There are some occasions where some phy nodes in the CI pool are
not properly cleaned up due to various reasons - hard crashes,
sudden network disconnects, etc. Whenever this happens, subsequent
job on the affected node usually fails due to overall instability
caused by lingering processes that should have been already killed.
Example of such failure is https://github.com/spdk/spdk/issues/2749.

This attempts to mitage potential impact of these issues under
per-patch testing.

Signed-off-by: Michal Berger <michal.berger@intel.com>
Change-Id: I6e5c834ab77813e7484fa7f65c9d39967a305329
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/16364
Reviewed-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
This commit is contained in:
Michal Berger 2023-01-20 01:29:21 +01:00 committed by Tomasz Zawadzki
parent 929ac2180b
commit 575a302d75

View File

@ -113,6 +113,10 @@ done
sync
if ! xtrace_disable_per_cmd reap_spdk_processes; then
echo "WARNING: Lingering SPDK processes were detected. Testing environment may be unstable" >&2
fi
if [ $(uname -s) = Linux ]; then
run_test "setup.sh" "$rootdir/test/setup/test-setup.sh"
fi