[2/2] test: static,poll -- Increase delta

Submitted by Cyrill Gorcunov on April 28, 2018, 9:56 a.m.

Details

Message ID 20180428095640.6460-3-gorcunov@gmail.com
State New
Series "zdtm: Add 'env' key"
Headers show

Commit Message

Cyrill Gorcunov April 28, 2018, 9:56 a.m.
The test case may consume more than 5 seconds depending
on the node load, for this sake we introduce ZDTM_DELTA
variable (see ae48b5d676e90e998cf6b12d3a36d5f73c45f5ea)
but actually there was no way to setup it in our vanilla
zdtm.py engine earlier. Now we can do that so setup the
limit up to one minute, thus if it trigger it is definitely
something goes wrong.

https://ci.openvz.org/view/CRIU/job/CRIU/job/CRIU-snap/job/criu-dev/3717/consoleFull
https://ci.openvz.org/view/CRIU/job/CRIU/job/CRIU-x86_64-pre-dump/job/criu-dev/3724/consoleFull

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>
---
 test/zdtm/static/poll.desc | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

Patch hide | download patch | download mbox

diff --git a/test/zdtm/static/poll.desc b/test/zdtm/static/poll.desc
index 63df42aa670a..f8bce6e5d61b 100644
--- a/test/zdtm/static/poll.desc
+++ b/test/zdtm/static/poll.desc
@@ -1 +1 @@ 
-{'flavor': 'h'}
+{'flavor': 'h', 'env': 'ZDTM_DELTA=60'}

Comments

Andrey Vagin Aug. 16, 2018, 11:59 p.m.
On Sat, Apr 28, 2018 at 12:56:40PM +0300, Cyrill Gorcunov wrote:
> The test case may consume more than 5 seconds depending
> on the node load, for this sake we introduce ZDTM_DELTA
> variable (see ae48b5d676e90e998cf6b12d3a36d5f73c45f5ea)
> but actually there was no way to setup it in our vanilla
> zdtm.py engine earlier. Now we can do that so setup the
> limit up to one minute, thus if it trigger it is definitely
> something goes wrong.
> 
> https://ci.openvz.org/view/CRIU/job/CRIU/job/CRIU-snap/job/criu-dev/3717/consoleFull
> https://ci.openvz.org/view/CRIU/job/CRIU/job/CRIU-x86_64-pre-dump/job/criu-dev/3724/consoleFull
> 
> Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>
> ---
>  test/zdtm/static/poll.desc | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/test/zdtm/static/poll.desc b/test/zdtm/static/poll.desc
> index 63df42aa670a..f8bce6e5d61b 100644
> --- a/test/zdtm/static/poll.desc
> +++ b/test/zdtm/static/poll.desc
> @@ -1 +1 @@
> -{'flavor': 'h'}
> +{'flavor': 'h', 'env': 'ZDTM_DELTA=60'}

gettimeofday(&time1, NULL)
poll(ufds, 2, delta * 1000);
gettimeofday(&time2, NULL)

if ((time2.tv_sec - time1.tv_sec) > delta) {
	fail("Delta is too big %lu",
		(unsigned long)(time2.tv_sec - time1.tv_sec));


I don't understand how this patch fixes the problem. It is obvious that
time2 - time1 is greater than delta. Maybe we need to take into account
some inaccuracy?

> -- 
> 2.14.3
>
Cyrill Gorcunov Aug. 28, 2018, 6:56 p.m.
On Fri, Aug 17, 2018 at 3:00 AM Andrey Vagin <avagin@virtuozzo.com> wrote:

> On Sat, Apr 28, 2018 at 12:56:40PM +0300, Cyrill Gorcunov wrote:
> > The test case may consume more than 5 seconds depending
> > on the node load, for this sake we introduce ZDTM_DELTA
> > variable (see ae48b5d676e90e998cf6b12d3a36d5f73c45f5ea)
> > but actually there was no way to setup it in our vanilla
> > zdtm.py engine earlier. Now we can do that so setup the
> > limit up to one minute, thus if it trigger it is definitely
> > something goes wrong.
> >
> >
> https://ci.openvz.org/view/CRIU/job/CRIU/job/CRIU-snap/job/criu-dev/3717/consoleFull
> >
> https://ci.openvz.org/view/CRIU/job/CRIU/job/CRIU-x86_64-pre-dump/job/criu-dev/3724/consoleFull
> >
> > Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>
> > ---
> >  test/zdtm/static/poll.desc | 2 +-
> >  1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > diff --git a/test/zdtm/static/poll.desc b/test/zdtm/static/poll.desc
> > index 63df42aa670a..f8bce6e5d61b 100644
> > --- a/test/zdtm/static/poll.desc
> > +++ b/test/zdtm/static/poll.desc
> > @@ -1 +1 @@
> > -{'flavor': 'h'}
> > +{'flavor': 'h', 'env': 'ZDTM_DELTA=60'}
>
> gettimeofday(&time1, NULL)
> poll(ufds, 2, delta * 1000);
> gettimeofday(&time2, NULL)
>
> if ((time2.tv_sec - time1.tv_sec) > delta) {
>         fail("Delta is too big %lu",
>                 (unsigned long)(time2.tv_sec - time1.tv_sec));
>
>
> I don't understand how this patch fixes the problem. It is obvious that
> time2 - time1 is greater than delta. Maybe we need to take into account
> some inaccuracy?
>

This is mostly for vz7 instance of testing where we have  time virtualized.
I'll review more detailed tomorrow and reply. Sorry for delay.