mesos-reviews mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Alexander Rukletsov <ruklet...@gmail.com>
Subject Re: Review Request 63368: Added MemoryProfiler class to Libprocess.
Date Wed, 28 Feb 2018 22:04:43 GMT

-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/63368/#review198409
-----------------------------------------------------------




3rdparty/libprocess/src/memory_profiler.cpp
Lines 305 (patched)
<https://reviews.apache.org/r/63368/#comment278541>

    This is a speculation, `result` should be informative enough. A likely reason is that
`jeprof` is absent on the machine, in which case this line is confusing.



3rdparty/libprocess/src/memory_profiler.cpp
Lines 312-322 (patched)
<https://reviews.apache.org/r/63368/#comment278542>

    Ideally we would respond with a simply formatted data, e.g., JSON, if a request comes
from `curl`/`httpie` and alike, and a "proper" html if it comes from a browser.
    
    Our help endpoints try to detect `curl` and `httpie` and respond accordingly. Another
approach would be to introduce a parameter to the endpoints, e.g., `format=json` that allows
to control the format of the response.



3rdparty/libprocess/src/memory_profiler.cpp
Lines 675 (patched)
<https://reviews.apache.org/r/63368/#comment278543>

    Can you please explain why you need default capture-by-reference here and below?



3rdparty/libprocess/src/memory_profiler.cpp
Lines 689 (patched)
<https://reviews.apache.org/r/63368/#comment278539>

    s\in"\in "



3rdparty/libprocess/src/memory_profiler.cpp
Lines 689-692 (patched)
<https://reviews.apache.org/r/63368/#comment278540>

    Technically, you do not redirect to the result, but you redirect to the `stop` endpoint
if the browser is used. `curl` users will have to request `stop` manually:
    ```
    alex@gru1: ~$ http 192.99.40.208:5555/memory-profiler/start?duration=1mins
    HTTP/1.1 200 OK
    Content-Length: 175
    Content-Type: text/html
    Date: Wed, 28 Feb 2018 21:32:42 GMT
    
    <html>  <head><meta http-equiv="refresh" content="60;url=./stop" /></head>
 <body>Started new heap profiling run.<br/>You will be redirected to the result
in60s.</body></html>
    ```


- Alexander Rukletsov


On Feb. 21, 2018, 10:41 a.m., Benno Evers wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/63368/
> -----------------------------------------------------------
> 
> (Updated Feb. 21, 2018, 10:41 a.m.)
> 
> 
> Review request for mesos, Alexander Rukletsov and Benjamin Mahler.
> 
> 
> Repository: mesos
> 
> 
> Description
> -------
> 
> This class exposes profiling functionality of jemalloc memory allocator
> when it is detected to be the memory allocator of the current process.
> 
> In particular, it gives developers an easy method to collect and
> access heap profiles which report which pieces of code were
> responsible for allocating memory.
> 
> 
> Diffs
> -----
> 
>   3rdparty/libprocess/Makefile.am c895d3ac7b9cc5ffd6c8b57ff24def866eb0213d 
>   3rdparty/libprocess/include/process/memory_profiler.hpp PRE-CREATION 
>   3rdparty/libprocess/include/process/process.hpp 8661706cb058efb26f5bfbcc84972f9930d3670f

>   3rdparty/libprocess/src/CMakeLists.txt 0ce7dac5deea94623530820d173ce3ffe5b42ea4 
>   3rdparty/libprocess/src/memory_profiler.cpp PRE-CREATION 
>   3rdparty/libprocess/src/process.cpp 42e7adf740b234ebf23d2dcb71440749c0ed87ec 
> 
> 
> Diff: https://reviews.apache.org/r/63368/diff/6/
> 
> 
> Testing
> -------
> 
> 
> Thanks,
> 
> Benno Evers
> 
>


Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message