DPDK patches and discussions
 help / color / mirror / Atom feed
* [dpdk-dev] [PATCH] ci: increase unit test timeout
@ 2020-01-28 16:28 Ferruh Yigit
  2020-01-28 18:39 ` Michael Santana
  2020-01-28 20:53 ` Aaron Conole
  0 siblings, 2 replies; 7+ messages in thread
From: Ferruh Yigit @ 2020-01-28 16:28 UTC (permalink / raw)
  To: Aaron Conole, Michael Santana; +Cc: dev

Timeout multiplier was 3, which gives 30 seconds for unit test but still
some unit test was timing out time to time and travis reporting false
positive failures.

Increasing the multiplier to 10, which makes timeout duration
100seconds.

Signed-off-by: Ferruh Yigit <ferruh.yigit@intel.com>
---
 .ci/linux-build.sh | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/.ci/linux-build.sh b/.ci/linux-build.sh
index ccc3a7ccd..be3dc4940 100755
--- a/.ci/linux-build.sh
+++ b/.ci/linux-build.sh
@@ -37,5 +37,5 @@ if [ "$AARCH64" != "1" ]; then
 fi
 
 if [ "$RUN_TESTS" = "1" ]; then
-    sudo meson test -C build --suite fast-tests -t 3
+    sudo meson test -C build --suite fast-tests -t 10
 fi
-- 
2.24.1


^ permalink raw reply	[flat|nested] 7+ messages in thread

* Re: [dpdk-dev] [PATCH] ci: increase unit test timeout
  2020-01-28 16:28 [dpdk-dev] [PATCH] ci: increase unit test timeout Ferruh Yigit
@ 2020-01-28 18:39 ` Michael Santana
  2020-01-28 20:53 ` Aaron Conole
  1 sibling, 0 replies; 7+ messages in thread
From: Michael Santana @ 2020-01-28 18:39 UTC (permalink / raw)
  To: Ferruh Yigit, Aaron Conole; +Cc: dev

I have been out of the loop for a while, so I do not know how many false 
positives there are. Looking at 
https://travis-ci.com/ovsrobot/dpdk/builds up to two weeks back it seems 
that hash_readwrite_lf_autotest is the only culprit. It looks like it 
lives in the neighborhood of 20-30s seconds so increasing the time out 
for this test seems appropriate as sometimes it halts at exactly 30s 
because of the time out.

Acked-by: Michael Santana <maicolgabriel@hotmail.com>

On 1/28/2020 11:28 AM, Ferruh Yigit wrote:
> Timeout multiplier was 3, which gives 30 seconds for unit test but still
> some unit test was timing out time to time and travis reporting false
> positive failures.
>
> Increasing the multiplier to 10, which makes timeout duration
> 100seconds.
>
> Signed-off-by: Ferruh Yigit <ferruh.yigit@intel.com>
> ---
>   .ci/linux-build.sh | 2 +-
>   1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/.ci/linux-build.sh b/.ci/linux-build.sh
> index ccc3a7ccd..be3dc4940 100755
> --- a/.ci/linux-build.sh
> +++ b/.ci/linux-build.sh
> @@ -37,5 +37,5 @@ if [ "$AARCH64" != "1" ]; then
>   fi
>   
>   if [ "$RUN_TESTS" = "1" ]; then
> -    sudo meson test -C build --suite fast-tests -t 3
> +    sudo meson test -C build --suite fast-tests -t 10
>   fi

^ permalink raw reply	[flat|nested] 7+ messages in thread

* Re: [dpdk-dev] [PATCH] ci: increase unit test timeout
  2020-01-28 16:28 [dpdk-dev] [PATCH] ci: increase unit test timeout Ferruh Yigit
  2020-01-28 18:39 ` Michael Santana
@ 2020-01-28 20:53 ` Aaron Conole
  2020-01-30 11:03   ` Thomas Monjalon
  1 sibling, 1 reply; 7+ messages in thread
From: Aaron Conole @ 2020-01-28 20:53 UTC (permalink / raw)
  To: Ferruh Yigit; +Cc: Aaron Conole, Michael Santana, dev

Ferruh Yigit <ferruh.yigit@intel.com> writes:

> Timeout multiplier was 3, which gives 30 seconds for unit test but still
> some unit test was timing out time to time and travis reporting false
> positive failures.
>
> Increasing the multiplier to 10, which makes timeout duration
> 100seconds.
>
> Signed-off-by: Ferruh Yigit <ferruh.yigit@intel.com>
> ---

It's okay to me.  I thought there was an effort to split out performance
part of this test from the functional part, but that seems to not have
gone anywhere.

Acked-by: Aaron Conole <aconole@redhat.com>


>  .ci/linux-build.sh | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/.ci/linux-build.sh b/.ci/linux-build.sh
> index ccc3a7ccd..be3dc4940 100755
> --- a/.ci/linux-build.sh
> +++ b/.ci/linux-build.sh
> @@ -37,5 +37,5 @@ if [ "$AARCH64" != "1" ]; then
>  fi
>  
>  if [ "$RUN_TESTS" = "1" ]; then
> -    sudo meson test -C build --suite fast-tests -t 3
> +    sudo meson test -C build --suite fast-tests -t 10
>  fi


^ permalink raw reply	[flat|nested] 7+ messages in thread

* Re: [dpdk-dev] [PATCH] ci: increase unit test timeout
  2020-01-28 20:53 ` Aaron Conole
@ 2020-01-30 11:03   ` Thomas Monjalon
  2020-01-30 15:35     ` Honnappa Nagarahalli
  2020-02-05 18:47     ` David Marchand
  0 siblings, 2 replies; 7+ messages in thread
From: Thomas Monjalon @ 2020-01-30 11:03 UTC (permalink / raw)
  To: Ferruh Yigit, Aaron Conole, Amit Gupta
  Cc: dev, Michael Santana, dev, Aaron Conole, david.marchand,
	yipeng1.wang, honnappa.nagarahalli

28/01/2020 21:53, Aaron Conole:
> Ferruh Yigit <ferruh.yigit@intel.com> writes:
> 
> > Timeout multiplier was 3, which gives 30 seconds for unit test but still
> > some unit test was timing out time to time and travis reporting false
> > positive failures.
> >
> > Increasing the multiplier to 10, which makes timeout duration
> > 100seconds.
> >
> > Signed-off-by: Ferruh Yigit <ferruh.yigit@intel.com>
> > ---
> 
> It's okay to me.  I thought there was an effort to split out performance
> part of this test from the functional part, but that seems to not have
> gone anywhere.
> 
> Acked-by: Aaron Conole <aconole@redhat.com>

NACK
The fix should be to split perf tests out of fast-tests.

The following patch is splitting hash_readwrite_autotest:
	https://patchwork.dpdk.org/patch/58726/
But we are still waiting for a patch splitting hash_readwrite_lf_autotest.
Please consider working on unit tests as a HIGH PRIORITY (using uppercase ;).
We should not have to wait so long to see performance tests removed
from fast unit tests (while keeping the functional coverage).




^ permalink raw reply	[flat|nested] 7+ messages in thread

* Re: [dpdk-dev] [PATCH] ci: increase unit test timeout
  2020-01-30 11:03   ` Thomas Monjalon
@ 2020-01-30 15:35     ` Honnappa Nagarahalli
  2020-01-31 15:44       ` Honnappa Nagarahalli
  2020-02-05 18:47     ` David Marchand
  1 sibling, 1 reply; 7+ messages in thread
From: Honnappa Nagarahalli @ 2020-01-30 15:35 UTC (permalink / raw)
  To: thomas, Ferruh Yigit, Aaron Conole, Amit Gupta
  Cc: dev, Michael Santana, dev, Aaron Conole, david.marchand,
	yipeng1.wang, Honnappa Nagarahalli, nd, nd



> -----Original Message-----
> From: Thomas Monjalon <thomas@monjalon.net>
> Sent: Thursday, January 30, 2020 5:03 AM
> To: Ferruh Yigit <ferruh.yigit@intel.com>; Aaron Conole
> <aconole@redhat.com>; Amit Gupta <agupta3@marvell.com>
> Cc: dev@dpdk.org; Michael Santana <maicolgabriel@hotmail.com>;
> dev@dpdk.org; Aaron Conole <aconole@redhat.com>;
> david.marchand@redhat.com; yipeng1.wang@intel.com; Honnappa
> Nagarahalli <Honnappa.Nagarahalli@arm.com>
> Subject: Re: [dpdk-dev] [PATCH] ci: increase unit test timeout
> 
> 28/01/2020 21:53, Aaron Conole:
> > Ferruh Yigit <ferruh.yigit@intel.com> writes:
> >
> > > Timeout multiplier was 3, which gives 30 seconds for unit test but
> > > still some unit test was timing out time to time and travis
> > > reporting false positive failures.
> > >
> > > Increasing the multiplier to 10, which makes timeout duration
> > > 100seconds.
> > >
> > > Signed-off-by: Ferruh Yigit <ferruh.yigit@intel.com>
> > > ---
> >
> > It's okay to me.  I thought there was an effort to split out
> > performance part of this test from the functional part, but that seems
> > to not have gone anywhere.
> >
> > Acked-by: Aaron Conole <aconole@redhat.com>
> 
> NACK
> The fix should be to split perf tests out of fast-tests.
> 
> The following patch is splitting hash_readwrite_autotest:
> 	https://patchwork.dpdk.org/patch/58726/
> But we are still waiting for a patch splitting hash_readwrite_lf_autotest.
> Please consider working on unit tests as a HIGH PRIORITY (using uppercase ;).
> We should not have to wait so long to see performance tests removed from
> fast unit tests (while keeping the functional coverage).
Apologies, it slipped from my radar. Will send out a patch soon.

> 
> 


^ permalink raw reply	[flat|nested] 7+ messages in thread

* Re: [dpdk-dev] [PATCH] ci: increase unit test timeout
  2020-01-30 15:35     ` Honnappa Nagarahalli
@ 2020-01-31 15:44       ` Honnappa Nagarahalli
  0 siblings, 0 replies; 7+ messages in thread
From: Honnappa Nagarahalli @ 2020-01-31 15:44 UTC (permalink / raw)
  To: thomas, Ferruh Yigit, Aaron Conole, Amit Gupta
  Cc: dev, Michael Santana, dev, Aaron Conole, david.marchand,
	yipeng1.wang, nd, Honnappa Nagarahalli, nd

> > Subject: Re: [dpdk-dev] [PATCH] ci: increase unit test timeout
> >
> > 28/01/2020 21:53, Aaron Conole:
> > > Ferruh Yigit <ferruh.yigit@intel.com> writes:
> > >
> > > > Timeout multiplier was 3, which gives 30 seconds for unit test but
> > > > still some unit test was timing out time to time and travis
> > > > reporting false positive failures.
> > > >
> > > > Increasing the multiplier to 10, which makes timeout duration
> > > > 100seconds.
> > > >
> > > > Signed-off-by: Ferruh Yigit <ferruh.yigit@intel.com>
> > > > ---
> > >
> > > It's okay to me.  I thought there was an effort to split out
> > > performance part of this test from the functional part, but that
> > > seems to not have gone anywhere.
> > >
> > > Acked-by: Aaron Conole <aconole@redhat.com>
> >
> > NACK
> > The fix should be to split perf tests out of fast-tests.
> >
> > The following patch is splitting hash_readwrite_autotest:
> > https://patchwork.dpdk.org/patch/58726/
Thomas/David, is it possible to merge this patch? It is acked already. It would be good to have these changes for LF functional tests.

> > But we are still waiting for a patch splitting hash_readwrite_lf_autotest.
> > Please consider working on unit tests as a HIGH PRIORITY (using uppercase ;).
> > We should not have to wait so long to see performance tests removed
> > from fast unit tests (while keeping the functional coverage).
> Apologies, it slipped from my radar. Will send out a patch soon.
> 
> >
> >
> 


^ permalink raw reply	[flat|nested] 7+ messages in thread

* Re: [dpdk-dev] [PATCH] ci: increase unit test timeout
  2020-01-30 11:03   ` Thomas Monjalon
  2020-01-30 15:35     ` Honnappa Nagarahalli
@ 2020-02-05 18:47     ` David Marchand
  1 sibling, 0 replies; 7+ messages in thread
From: David Marchand @ 2020-02-05 18:47 UTC (permalink / raw)
  To: Thomas Monjalon
  Cc: Ferruh Yigit, Aaron Conole, Amit Gupta, dev, Michael Santana,
	Wang, Yipeng1, Honnappa Nagarahalli

On Thu, Jan 30, 2020 at 12:03 PM Thomas Monjalon <thomas@monjalon.net> wrote:
>
> 28/01/2020 21:53, Aaron Conole:
> > Ferruh Yigit <ferruh.yigit@intel.com> writes:
> >
> > > Timeout multiplier was 3, which gives 30 seconds for unit test but still
> > > some unit test was timing out time to time and travis reporting false
> > > positive failures.
> > >
> > > Increasing the multiplier to 10, which makes timeout duration
> > > 100seconds.
> > >
> > > Signed-off-by: Ferruh Yigit <ferruh.yigit@intel.com>
> > > ---
> >
> > It's okay to me.  I thought there was an effort to split out performance
> > part of this test from the functional part, but that seems to not have
> > gone anywhere.
> >
> > Acked-by: Aaron Conole <aconole@redhat.com>
>
> NACK
> The fix should be to split perf tests out of fast-tests.
>
> The following patch is splitting hash_readwrite_autotest:
>         https://patchwork.dpdk.org/patch/58726/
> But we are still waiting for a patch splitting hash_readwrite_lf_autotest.
> Please consider working on unit tests as a HIGH PRIORITY (using uppercase ;).
> We should not have to wait so long to see performance tests removed
> from fast unit tests (while keeping the functional coverage).

Just applied series
https://patchwork.dpdk.org/project/dpdk/list/?series=8401&state=*.
So marking this patch as rejected.

Thanks.


-- 
David Marchand


^ permalink raw reply	[flat|nested] 7+ messages in thread

end of thread, other threads:[~2020-02-05 18:47 UTC | newest]

Thread overview: 7+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2020-01-28 16:28 [dpdk-dev] [PATCH] ci: increase unit test timeout Ferruh Yigit
2020-01-28 18:39 ` Michael Santana
2020-01-28 20:53 ` Aaron Conole
2020-01-30 11:03   ` Thomas Monjalon
2020-01-30 15:35     ` Honnappa Nagarahalli
2020-01-31 15:44       ` Honnappa Nagarahalli
2020-02-05 18:47     ` David Marchand

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).