On Thu, Oct 24, 2013 at 7:01 AM, Stanislav Fomichev <stfomic...@yandex-team.ru> wrote: >> I feel very nervous with this one. Let's say the setup is for cpu, >> with nested groups such a and a/b. What if something fails to be set >> in a, or a/b which then causes problems with containment/other groups. >> You would be asking for the user to check whether cgconfigparser >> actually succeeded or not. That sounds like a regression to me. > Currently we have the following situation: > call to cgconfigparser is buried somewhere in init scripts and if it fails > setting some group variable we get no cgroups at all. > User doesn't know about it without checking logs or cgroups setup. > > What I'd like to have: cgconfigparser should try to set as much as > possible and report errors to the log, so we get almost functional > cgroups setup (in my case I'd like to share one cgconfig between different > kernels which may miss some group variables). > > For me it's better to have cgroups running with some values missing, > than to have no cgroups at all. As a compromise we may add some flag to > the cgconfigparser which tells it to be strict and treat all errors as fatal > errors to preserve the old behavior for the users that still need it.
I would prefer a flag for your behaviour. Thanks! Dhaval ------------------------------------------------------------------------------ October Webinars: Code for Performance Free Intel webinars can help you accelerate application performance. Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most from the latest Intel processors and coprocessors. See abstracts and register > http://pubads.g.doubleclick.net/gampad/clk?id=60135991&iu=/4140/ostg.clktrk _______________________________________________ Libcg-devel mailing list Libcg-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/libcg-devel
