[CRIU] The --check-only feature

Andrey Vagin avagin at virtuozzo.com
Fri Sep 22 02:52:46 MSK 2017


On Thu, Sep 21, 2017 at 05:51:09PM +0300, Pavel Emelyanov wrote:
> On 09/21/2017 05:18 PM, Adrian Reber wrote:
> > On Thu, Sep 21, 2017 at 04:35:31PM +0300, Pavel Emelyanov wrote:
> >> We've had the --check-only patches in criu-dev for a while. What do you
> >> think, is the feature ready to get merged into master branch?
> > 
> > I have not looked at it for a while, but Andrey did some changes in the
> > past so it seems to be working.
> > 
> >> There's a checklist for every new feature: https://criu.org/New_features_checklist
> >> Do we have everything (applicable) for the --check-only?
> > 
> > It is documented on the wiki and in the man-page.
> 
> Cool
> 
> > It is integrated into zdtm, but I am not sure if it used in
> 
> BTW, about zdtm, are all tests --check-only-ed, or is there some excusion list?
> 
> > jenkins/travis (I actually don't know the difference).
> 
> On travis we run some tests (not all), check compilation for all the arches and
> do some other sanity checks. This is run on every single patch(set) that hits
> the ML. On jenkins we have a wider range of tests, but only for x86 (and maybe
> ppc).
> 
> Andrey, where should we put the --check-only testing? I think travis would be a
> good fit, breaking this more would be easy.


We need to run all tests in Jenkins and a "small" set with maximum code
coverage in Travis.

> 
> > It is only available through CLI and not via RPC or the library.
> > 
> > So maybe we should get it in the normal test runs and try to merge in
> > master with 3.6.
> 
> Yes, 3.6 would be a good fit. May I ask you to make a patch that enables in for
> RPC? I think --check-only restore would be useful not just as CLI (i.e. -- a toy 
> for devs).
> 
> -- Pavel


More information about the CRIU mailing list