Commit 95e05721 authored by Monica Kenguva's avatar Monica Kenguva Committed by Tomasz Zawadzki
Browse files

test/nvmf: test reconnect_delay_sec parameter 



Signed-off-by: default avatarMonica Kenguva <monica.kenguva@intel.com>
Change-Id: I45dc2b2fe660e5a53c8976dea2640ea53ec00a3d
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/12184


Community-CI: Broadcom CI <spdk-ci.pdl@broadcom.com>
Community-CI: Mellanox Build Bot
Tested-by: default avatarSPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: default avatarJim Harris <james.r.harris@intel.com>
Reviewed-by: default avatarShuhei Matsumoto <smatsumoto@nvidia.com>
parent 617a7303

module/bdev/nvme/bdev_nvme.c

+3 −0
Original line number Diff line number Diff line
@@ -1591,6 +1591,7 @@ bdev_nvme_reconnect_delay_timer_expired(void *ctx) 
{

	struct nvme_ctrlr *nvme_ctrlr = ctx;



	SPDK_DTRACE_PROBE1(bdev_nvme_ctrlr_reconnect_delay, nvme_ctrlr->nbdev_ctrlr->name);

	pthread_mutex_lock(&nvme_ctrlr->mutex);



	spdk_poller_unregister(&nvme_ctrlr->reconnect_delay_timer);
@@ -1783,6 +1784,7 @@ bdev_nvme_reconnect_ctrlr(struct nvme_ctrlr *nvme_ctrlr) 
{

	spdk_nvme_ctrlr_reconnect_async(nvme_ctrlr->ctrlr);



	SPDK_DTRACE_PROBE1(bdev_nvme_ctrlr_reconnect, nvme_ctrlr->nbdev_ctrlr->name);

	assert(nvme_ctrlr->reset_detach_poller == NULL);

	nvme_ctrlr->reset_detach_poller = SPDK_POLLER_REGISTER(bdev_nvme_reconnect_ctrlr_poll,

					  nvme_ctrlr, 0);
@@ -1793,6 +1795,7 @@ bdev_nvme_reset_ctrlr(struct spdk_io_channel_iter *i, int status) 
{

	struct nvme_ctrlr *nvme_ctrlr = spdk_io_channel_iter_get_io_device(i);



	SPDK_DTRACE_PROBE1(bdev_nvme_ctrlr_reset, nvme_ctrlr->nbdev_ctrlr->name);

	assert(status == 0);



	if (!spdk_nvme_ctrlr_is_fabrics(nvme_ctrlr->ctrlr)) {

scripts/bpf/nvmf_timeout.bt

0 → 100644
+17 −0
Original line number Diff line number Diff line 
usdt:__EXE__:bdev_nvme_ctrlr_reconnect {

	printf("%d.%06d: reconnect bdev controller %s\n",

	       elapsed / (uint64)(1000 * 1000), elapsed % (uint64)(1000 * 1000),

	       str(arg1));

}



usdt:__EXE__:bdev_nvme_ctrlr_reconnect_delay {

	printf("%d.%06d: reconnect delay bdev controller %s\n",

	       elapsed / (uint64)(1000 * 1000), elapsed % (uint64)(1000 * 1000),

	       str(arg1));

}



usdt:__EXE__:bdev_nvme_ctrlr_reset {

	printf("%d.%06d: reset bdev controller %s\n",

	       elapsed / (uint64)(1000 * 1000), elapsed % (uint64)(1000 * 1000),

	       str(arg1));

}

test/nvmf/host/timeout.sh

+43 −1
Original line number Diff line number Diff line
@@ -9,6 +9,7 @@ MALLOC_BDEV_SIZE=64 
MALLOC_BLOCK_SIZE=512



rpc_py="$rootdir/scripts/rpc.py"

bpf_sh="$rootdir/scripts/bpftrace.sh"



bdevperf_rpc_sock=/var/tmp/bdevperf.sock
@@ -36,7 +37,7 @@ function get_controller() { 
	$rpc_py -s $bdevperf_rpc_sock bdev_nvme_get_controllers | jq -r '.[].name'

}



# case 1 test ctrlr_loss_timeout_sec time to try reconnecting to a ctrlr before deleting it

# Case 1 test ctrlr_loss_timeout_sec time to try reconnecting to a ctrlr before deleting it

# ctrlr_loss_timeout_sec is 10 reconnect_delay_sec is 5

$rpc_py -s $bdevperf_rpc_sock bdev_nvme_set_options -r -1

$rpc_py -s $bdevperf_rpc_sock bdev_nvme_attach_controller -b NVMe0 -t $TEST_TRANSPORT -a $NVMF_FIRST_TARGET_IP -s $NVMF_PORT -f ipv4 -n nqn.2016-06.io.spdk:cnode1 -l 10 -o 5
@@ -90,6 +91,8 @@ wait $rpc_pid 
cat $testdir/try.txt

cat /dev/null > $testdir/try.txt



# TODO: Check the IO fail if we wait for 5 sec, needs information from bdevperf



$rootdir/test/bdev/bdevperf/bdevperf.py -s $bdevperf_rpc_sock perform_tests &

rpc_pid=$!

sleep 1
@@ -103,8 +106,47 @@ cat $testdir/try.txt 
killprocess $bdevperf_pid

rm -f $testdir/try.txt



# Case 3 test reconnect_delay_sec

# Time to delay a reconnect trial

$rootdir/test/bdev/bdevperf/bdevperf -m 0x4 -z -r $bdevperf_rpc_sock -q 128 -o 4096 -w verify -t 20 -f &> $testdir/try.txt &

bdevperf_pid=$!



trap 'process_shm --id $NVMF_APP_SHM_ID; rm -f $testdir/try.txt; killprocess $bdevperf_pid; nvmftestfini; exit 1' SIGINT SIGTERM EXIT

waitforlisten $bdevperf_pid $bdevperf_rpc_sock



#start_trace

$bpf_sh $bdevperf_pid $rootdir/scripts/bpf/nvmf_timeout.bt &> $testdir/trace.txt &

dtrace_pid=$!



$rpc_py -s $bdevperf_rpc_sock bdev_nvme_set_options -r -1



# ctrlr_loss_timeout_sec is 10 reconnect_delay_sec is 2

$rpc_py -s $bdevperf_rpc_sock bdev_nvme_attach_controller -b NVMe0 -t $TEST_TRANSPORT -a $NVMF_FIRST_TARGET_IP \

	-s $NVMF_PORT -f ipv4 -n nqn.2016-06.io.spdk:cnode1 -l 10 -o 2

$rootdir/test/bdev/bdevperf/bdevperf.py -s $bdevperf_rpc_sock perform_tests &

rpc_pid=$!

sleep 1

$rpc_py nvmf_subsystem_remove_listener nqn.2016-06.io.spdk:cnode1 -t $TEST_TRANSPORT -a $NVMF_FIRST_TARGET_IP -s $NVMF_PORT



# wait for 5sec to count the reconnect delay frequency with 2 sec interval

sleep 5

cat $testdir/trace.txt



# Check the frequency of delay reconnect

if (("$(grep -c "reconnect delay bdev controller NVMe0" < $testdir/trace.txt)" != 2)); then

	false

fi



kill $dtrace_pid

rm -f $testdir/trace.txt

wait $rpc_pid

cat $testdir/try.txt



killprocess $bdevperf_pid



$rpc_py nvmf_delete_subsystem nqn.2016-06.io.spdk:cnode1



trap - SIGINT SIGTERM EXIT

rm -f $testdir/try.txt



nvmftestfini

test/nvmf/nvmf.sh

+3 −1
Original line number Diff line number Diff line
@@ -84,7 +84,9 @@ if [[ $SPDK_TEST_USDT -eq 1 ]]; then 
	run_test "nvmf_multipath" test/nvmf/host/multipath.sh "${TEST_ARGS[@]}"

fi

run_test "nvmf_discovery" test/nvmf/host/discovery.sh "${TEST_ARGS[@]}"

if [[ $SPDK_TEST_USDT -eq 1 ]]; then

	run_test "nvmf_timeout" test/nvmf/host/timeout.sh "${TEST_ARGS[@]}"

fi



# TODO: disabled due to intermittent failures (RDMA_CM_EVENT_UNREACHABLE/ETIMEDOUT)

#run_test test/nvmf/host/identify_kernel_nvmf.sh $TEST_ARGS