On Fri, Apr 12, 2019 at 5:36 PM Michal Hocko <mhocko@xxxxxxxxxx> wrote: > > On Fri 12-04-19 17:29:04, Yafang Shao wrote: > > On Fri, Apr 12, 2019 at 5:09 PM Michal Hocko <mhocko@xxxxxxxxxx> wrote: > [...] > > > > Then we can do some trace for this memcg, i.e. to trace how long the > > > > applicatons may stall via tracepoint. > > > > (but current tracepoints can't trace a specified cgroup only, that's > > > > another point to be improved.) > > > > > > It is a task that is stalled, not a cgroup. > > > > But these tracepoints can't filter a speficied task neither. > > each trace line output should cotain a pid, no? But that's not enough. Some drawbacks, - the PID is variable, and it is not so conveninet to get the tasks from this PID. i.e. when you use pidof to get the tasks, it may already exit and you get nothing. - the traceline don't always contain the task names. - if we don't filter the tasks with tracepoint filter, there may be lots of output. i.e. we always deploy lots of cgroup on a single host, but only some of them are important, while the others are not import. So we limit the not important cgroup to a low memory limit, and then the tasks in it may do frequent memcg reclaim, but we don't care. Thanks Yafang
