> On Feb. 3, 2016, 5:11 p.m., Jiang Yan Xu wrote: > > 3rdparty/libprocess/src/process.cpp, lines 2207-2208 > > <https://reviews.apache.org/r/43149/diff/2/?file=1231902#file1231902line2207> > > > > Since we do LOG(WARNING) instead of LOG(FATAL), maybe say the below so > > it's clear that we ignored the variable and moved on with the default value. > > > > `"Ignoring invalid value " << value.get() << " for > > LIBPROCESS_THREAD_COUNT";` > > Guangya Liu wrote: > What about adding more info for trouble shooting? Such as `"Ignoring > invalid value " << value.get() << " for LIBPROCESS_THREAD_COUNT, using system > defined value " << cpus;` > > haosdent huang wrote: > +1
Hmm... either this or we just EXIT(EXIT_FAILURE). I am leaning towards the latter after thinking about it a bit more. It's also more consistent with practices in this file: e.g. when LIBPROCESS_PORT is invalid we don't use the default port (and elsewhere in the project). > On Feb. 3, 2016, 5:11 p.m., Jiang Yan Xu wrote: > > 3rdparty/libprocess/src/process.cpp, line 2203 > > <https://reviews.apache.org/r/43149/diff/2/?file=1231902#file1231902line2203> > > > > `Try<size_t> count = numify<size_t>(...)` since the thread count is an > > uint? My bad for suggesting using numify<size_t>. Seems like numify<size_t> actually is going to underflow for a negative long and output a large number. So we ought to keep using long and do boundary checks explicitly. i.e. count.get() > 0 - Jiang Yan ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/43149/#review117725 ----------------------------------------------------------- On Feb. 3, 2016, 4:40 p.m., James Peach wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/43149/ > ----------------------------------------------------------- > > (Updated Feb. 3, 2016, 4:40 p.m.) > > > Review request for mesos, Benjamin Hindman and Jiang Yan Xu. > > > Bugs: MESOS-4592 > https://issues.apache.org/jira/browse/MESOS-4592 > > > Repository: mesos > > > Description > ------- > > Check the LIBPROCESS_THREAD_COUNT environment variable to determine > the number of threads in the libprocess thread pool. > > > Diffs > ----- > > 3rdparty/libprocess/src/process.cpp > 65a247a5bc7cf0bd6f23f6fc5c349ecce60f5ec0 > > Diff: https://reviews.apache.org/r/43149/diff/ > > > Testing > ------- > > make check on OS X. Running in production for many months. > > > Thanks, > > James Peach > >
