-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/32805/#review78801
-----------------------------------------------------------

Ship it!



src/linux/perf.cpp
<https://reviews.apache.org/r/32805/#comment127831>

    Add a comment here ?



src/linux/perf.cpp
<https://reviews.apache.org/r/32805/#comment127833>

    Definitely needs a comment here on what you are trying to do.



src/linux/perf.cpp
<https://reviews.apache.org/r/32805/#comment127840>

    Comment here that this child runs the perf command.



src/linux/perf.cpp
<https://reviews.apache.org/r/32805/#comment127842>

    Comment that the parent blocks until the child (i.e., perf command) exits.



src/linux/perf.cpp
<https://reviews.apache.org/r/32805/#comment127834>

    Comment here too.


- Vinod Kone


On April 3, 2015, 5:11 p.m., Jie Yu wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/32805/
> -----------------------------------------------------------
> 
> (Updated April 3, 2015, 5:11 p.m.)
> 
> 
> Review request for mesos, Ben Mahler, Ian Downes, and Vinod Kone.
> 
> 
> Bugs: MESOS-2462
>     https://issues.apache.org/jira/browse/MESOS-2462
> 
> 
> Repository: mesos
> 
> 
> Description
> -------
> 
> Terminated the perf subprocess once the parent exits.
> 
> The idea is to have a nanny process which installs a SIGTERM handler which 
> will kill the process group (of course, we need to put the nanny and perf 
> process to the same process group). We set the death signal of the nanny 
> process to be SIGTERM. In that way, when slave exits, the nanny process will 
> receive a SIGTERM, which will then kill all processes in the process group.
> 
> 
> Diffs
> -----
> 
>   src/linux/perf.cpp cad6c80e2a9608ff02fc2b8976efba52713dd5a8 
> 
> Diff: https://reviews.apache.org/r/32805/diff/
> 
> 
> Testing
> -------
> 
> sudo make check
> 
> I also manually verified it by terminating the slave while perf is in 
> progress. The perf is killed immediately.
> 
> 
> Thanks,
> 
> Jie Yu
> 
>

Reply via email to