test/nvmf: try to make reactors more busy in interrupt test

This test verifies that reactors are idle when there's no IO and that
they're idle when handling IO. The latter check can be a problem,
b

test/nvmf: try to make reactors more busy in interrupt test

This test verifies that reactors are idle when there's no IO and that
they're idle when handling IO. The latter check can be a problem,
because the business of a core when handling IOs varies and can never
reach the threshold that we use to consider a core to be busy.

To make that second check more reliable, let's do a couple of things:
increase queue depth to 256 and decrease busy threshold to 30. Also,
while here, use double parenthesis for arithmetic comparisons.

Fixes: #3520

Signed-off-by: Konrad Sztyber <konrad.sztyber@intel.com>
Change-Id: I2937b46471d523bb496e7b3878f43c541047adaf
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/24827
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@nvidia.com>
Reviewed-by: Jim Harris <jim.harris@samsung.com>

show more ...

test: make reactor_is_busy_or_idle() less restrictive

Extensive interrupt mode testing shows that reactor CPU load as reported
by the top in reactor_is_busy_or_idle during traffic fluctuates between

test: make reactor_is_busy_or_idle() less restrictive

Extensive interrupt mode testing shows that reactor CPU load as reported
by the top in reactor_is_busy_or_idle during traffic fluctuates between
50 and 80%, 66.7% is reported quite often. With 10s time slot it happens
that all readings are 66.7% or less.
To avoid false positives in CI and not make the test any longer than it
already is lower the arbitrary chosen value o 70% to 65% as an
indication that the CPU is indeed busy.

Change-Id: I4946ab2fc0b0704ef6e3d6905cfcae6331e8ec9c
Signed-off-by: Krzysztof Goreczny <krzysztof.goreczny@dell.com>
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/24596
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Konrad Sztyber <konrad.sztyber@intel.com>
Reviewed-by: Jim Harris <jim.harris@samsung.com>

show more ...

test: split interrupt_common.sh

This allows for reuse of truly common part of this file in upcoming
commit that introduce NVMe-of TCP interrupt mode support.

Change-Id: If28935aa21d71a9e97d89517206

test: split interrupt_common.sh

This allows for reuse of truly common part of this file in upcoming
commit that introduce NVMe-of TCP interrupt mode support.

Change-Id: If28935aa21d71a9e97d89517206c0a4f12dc21eb
Signed-off-by: Krzysztof Goreczny <krzysztof.goreczny@dell.com>
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/22835
Reviewed-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
Reviewed-by: Boris Glimcher <Boris.Glimcher@emc.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Jim Harris <jim.harris@samsung.com>

show more ...