LU-16652 tests: sanity-lnet test 253/254 race with discovery

The delay rule used for test 253/254 can race with incoming discovery
PUSH (PUT + ACK). This results in an extra GET being queued on the
peer NI tx queue and subsequently dropped.

Fix this issue by restricting the delay rule to GET message type.

Test-Parameters: trivial testlist=sanity-lnet env=ONLY=253,254
Signed-off-by: Chris Horn <chris.horn@hpe.com>
Change-Id: Ie66cc5e00cc3718e74221b080e099958cef7a401
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/50469
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Reviewed-by: Serguei Smirnov <ssmirnov@whamcloud.com>
Reviewed-by: Cyril Bordage <cbordage@whamcloud.com>
Reviewed-by: Frank Sehr <fsehr@whamcloud.com>
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>

diff --git a/lustre/tests/sanity-lnet.sh b/lustre/tests/sanity-lnet.sh
index b687058..939c629 100755 (executable)
--- a/lustre/tests/sanity-lnet.sh
+++ b/lustre/tests/sanity-lnet.sh
@@ -3252,7 +3252,8 @@ do_expired_message_drop_test() {
 
        for lnid in "${LNIDS[@]}"; do
                for rnid in "${RNIDS[@]}"; do
-                       $LCTL net_delay_add -s "${lnid}" -d "${rnid}" -l "${delay}" -r 1
+                       $LCTL net_delay_add -s "${lnid}" -d "${rnid}" \
+                               -l "${delay}" -r 1 -m GET
                done
        done