You are viewing a plain text version of this content. The canonical link for it is here.
Posted to reviews@mesos.apache.org by Maged Michael <ma...@gmail.com> on 2016/03/28 22:28:39 UTC
Re: Review Request 43144: Added env var to set upper bound on number
of libprocess worker threads.
-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/43144/
-----------------------------------------------------------
(Updated March 28, 2016, 8:28 p.m.)
Review request for mesos, Guangya Liu, Klaus Ma, and Qian Zhang.
Bugs: MESOS-4353
https://issues.apache.org/jira/browse/MESOS-4353
Repository: mesos
Description
-------
Added env var to set upper bound on number of libprocess worker threads.
Diffs (updated)
-----
3rdparty/libprocess/src/process.cpp feaffa4
docs/configuration.md 9ad0c2a
Diff: https://reviews.apache.org/r/43144/diff/
Testing
-------
Passed make check on x86_64 Ubuntu 14.04.
Invocations of mesos-master, mesos-slave, and mesos-tests with the LIBPROCESS_MAX_WORKER_THREADS set to the values "1", "7", "9", "1000", "0", "-1", and "abc" on a 4-core x86_64 Ubuntu 14.04 system. The results were as expected. 1, 7, 8, 8, 8, 8, and 8 worker threads were created, respectively. The last 3 tests generated warnings.
Thanks,
Maged Michael
Re: Review Request 43144: Added env var to set upper bound on number
of libprocess worker threads.
Posted by Maged Michael <ma...@gmail.com>.
-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/43144/
-----------------------------------------------------------
(Updated March 29, 2016, 11:14 p.m.)
Review request for mesos, Guangya Liu, Klaus Ma, and Qian Zhang.
Bugs: MESOS-4353
https://issues.apache.org/jira/browse/MESOS-4353
Repository: mesos
Description
-------
Added env var to set upper bound on number of libprocess worker threads.
Diffs (updated)
-----
3rdparty/libprocess/src/process.cpp feaffa4
docs/configuration.md 9ad0c2a
Diff: https://reviews.apache.org/r/43144/diff/
Testing
-------
Passed make check on x86_64 Ubuntu 14.04.
Invocations of mesos-master, mesos-slave, and mesos-tests with the LIBPROCESS_MAX_WORKER_THREADS set to the values "1", "7", "9", "1000", "0", "-1", and "abc" on a 4-core x86_64 Ubuntu 14.04 system. The results were as expected. 1, 7, 8, 8, 8, 8, and 8 worker threads were created, respectively. The last 3 tests generated warnings.
Thanks,
Maged Michael
Re: Review Request 43144: Added env var to set upper bound on number
of libprocess worker threads.
Posted by Maged Michael <ma...@gmail.com>.
> On March 29, 2016, 12:52 p.m., Joris Van Remoortere wrote:
> > 3rdparty/libprocess/src/process.cpp, line 2193
> > <https://reviews.apache.org/r/43144/diff/7/?file=1316988#file1316988line2193>
> >
> > since this value will not longer always map to the number of `cpus`, can we rename this to something more meaningful such as `num_worker_threads`?
Done. I also changed `cpus` in lines 846, 986 and 987.
> On March 29, 2016, 12:52 p.m., Joris Van Remoortere wrote:
> > 3rdparty/libprocess/src/process.cpp, line 2208
> > <https://reviews.apache.org/r/43144/diff/7/?file=1316988#file1316988line2208>
> >
> > please use full words for variable names (i.e. `num`)
Changed to `number`.
> On March 29, 2016, 12:52 p.m., Joris Van Remoortere wrote:
> > 3rdparty/libprocess/src/process.cpp, lines 2210-2212
> > <https://reviews.apache.org/r/43144/diff/7/?file=1316988#file1316988line2210>
> >
> > It seems valuable to log the number of threads we will use regardless of whether they were over-ridden or not.
> >
> > Let's log the actual number outside of the `if` block. This will also simplify the message (e.g. `Using X libprocess worker threads`
The number of thread is logged in lines 986 and 987. The added logging here is just for the override.
What do you think?
> On March 29, 2016, 12:52 p.m., Joris Van Remoortere wrote:
> > docs/configuration.md, line 1761
> > <https://reviews.apache.org/r/43144/diff/7/?file=1316989#file1316989line1761>
> >
> > Shall we explain what the default is?
Added ", which is the maximum of 8 and the number of cores on the machine".
- Maged
-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/43144/#review125864
-----------------------------------------------------------
On March 29, 2016, 11:14 p.m., Maged Michael wrote:
>
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/43144/
> -----------------------------------------------------------
>
> (Updated March 29, 2016, 11:14 p.m.)
>
>
> Review request for mesos, Guangya Liu, Klaus Ma, and Qian Zhang.
>
>
> Bugs: MESOS-4353
> https://issues.apache.org/jira/browse/MESOS-4353
>
>
> Repository: mesos
>
>
> Description
> -------
>
> Added env var to set upper bound on number of libprocess worker threads.
>
>
> Diffs
> -----
>
> 3rdparty/libprocess/src/process.cpp feaffa4
> docs/configuration.md 9ad0c2a
>
> Diff: https://reviews.apache.org/r/43144/diff/
>
>
> Testing
> -------
>
> Passed make check on x86_64 Ubuntu 14.04.
>
> Invocations of mesos-master, mesos-slave, and mesos-tests with the LIBPROCESS_MAX_WORKER_THREADS set to the values "1", "7", "9", "1000", "0", "-1", and "abc" on a 4-core x86_64 Ubuntu 14.04 system. The results were as expected. 1, 7, 8, 8, 8, 8, and 8 worker threads were created, respectively. The last 3 tests generated warnings.
>
>
> Thanks,
>
> Maged Michael
>
>
Re: Review Request 43144: Added env var to set upper bound on number
of libprocess worker threads.
Posted by Joris Van Remoortere <jo...@gmail.com>.
-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/43144/#review125864
-----------------------------------------------------------
3rdparty/libprocess/src/process.cpp (line 2193)
<https://reviews.apache.org/r/43144/#comment188744>
since this value will not longer always map to the number of `cpus`, can we rename this to something more meaningful such as `num_worker_threads`?
3rdparty/libprocess/src/process.cpp (line 2204)
<https://reviews.apache.org/r/43144/#comment188745>
I don't think we tend to use `auto` for this pattern. Do you mind explicitly stating the type?
3rdparty/libprocess/src/process.cpp (line 2208)
<https://reviews.apache.org/r/43144/#comment188746>
please use full words for variable names (i.e. `num`)
3rdparty/libprocess/src/process.cpp (lines 2210 - 2212)
<https://reviews.apache.org/r/43144/#comment188743>
It seems valuable to log the number of threads we will use regardless of whether they were over-ridden or not.
Let's log the actual number outside of the `if` block. This will also simplify the message (e.g. `Using X libprocess worker threads`
docs/configuration.md (line 1761)
<https://reviews.apache.org/r/43144/#comment188747>
Shall we explain what the default is?
- Joris Van Remoortere
On March 28, 2016, 8:28 p.m., Maged Michael wrote:
>
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/43144/
> -----------------------------------------------------------
>
> (Updated March 28, 2016, 8:28 p.m.)
>
>
> Review request for mesos, Guangya Liu, Klaus Ma, and Qian Zhang.
>
>
> Bugs: MESOS-4353
> https://issues.apache.org/jira/browse/MESOS-4353
>
>
> Repository: mesos
>
>
> Description
> -------
>
> Added env var to set upper bound on number of libprocess worker threads.
>
>
> Diffs
> -----
>
> 3rdparty/libprocess/src/process.cpp feaffa4
> docs/configuration.md 9ad0c2a
>
> Diff: https://reviews.apache.org/r/43144/diff/
>
>
> Testing
> -------
>
> Passed make check on x86_64 Ubuntu 14.04.
>
> Invocations of mesos-master, mesos-slave, and mesos-tests with the LIBPROCESS_MAX_WORKER_THREADS set to the values "1", "7", "9", "1000", "0", "-1", and "abc" on a 4-core x86_64 Ubuntu 14.04 system. The results were as expected. 1, 7, 8, 8, 8, 8, and 8 worker threads were created, respectively. The last 3 tests generated warnings.
>
>
> Thanks,
>
> Maged Michael
>
>