mesos-reviews mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Jacob Janco <jjanco....@gmail.com>
Subject Re: Review Request 53479: Perform agent GC asynchronously.
Date Sat, 15 Jul 2017 01:20:35 GMT


> On July 14, 2017, 8:47 a.m., Jiang Yan Xu wrote:
> > src/slave/gc.hpp
> > Lines 122-123 (patched)
> > <https://reviews.apache.org/r/53479/diff/6/?file=1776250#file1776250line123>
> >
> >     No need for pointer?
> >     
> >     Just `process::Promise<Nothing> promise;` which is default initialized.

My mistake, I added an overload to Promise to help with the PathInfo check since we're not
checking the pointer anymore.


> On July 14, 2017, 8:47 a.m., Jiang Yan Xu wrote:
> > src/slave/gc.cpp
> > Lines 150-152 (patched)
> > <https://reviews.apache.org/r/53479/diff/6/?file=1776251#file1776251line156>
> >
> >     This comment doesn't seem necessary, we may have missed this initially but retrospectively
resetting `removing` before starting async rmdir feels straighforward.

Cool, removing.


> On July 14, 2017, 8:47 a.m., Jiang Yan Xu wrote:
> > src/slave/gc.cpp
> > Lines 155 (patched)
> > <https://reviews.apache.org/r/53479/diff/6/?file=1776251#file1776251line161>
> >
> >     Also add the action taken here, which is 
> >     
> >     ```
> >     VLOG(1) << "Skipping deletion of '" << info-> path << "'
 as it is already in progress";
> >     ```

Fixed.


> On July 14, 2017, 8:47 a.m., Jiang Yan Xu wrote:
> > src/slave/gc.cpp
> > Lines 159 (patched)
> > <https://reviews.apache.org/r/53479/diff/6/?file=1776251#file1776251line165>
> >
> >     Not yours but quoting the path makes logging more consistent.
> >     
> >     Also move this into the lambda above each rmdir because this log does suggest
the immediate rmdir for this path?

That makes more sense, moving the log line there.


> On July 14, 2017, 8:47 a.m., Jiang Yan Xu wrote:
> > src/slave/gc.cpp
> > Lines 163 (patched)
> > <https://reviews.apache.org/r/53479/diff/6/?file=1776251#file1776251line169>
> >
> >     I think the name and type `bool removing` is self-evident.

Removed comment.


> On July 14, 2017, 8:47 a.m., Jiang Yan Xu wrote:
> > src/slave/gc.cpp
> > Lines 205 (patched)
> > <https://reviews.apache.org/r/53479/diff/6/?file=1776251#file1776251line213>
> >
> >     `CHECK_READY(result);`?

Added.


> On July 14, 2017, 8:47 a.m., Jiang Yan Xu wrote:
> > src/slave/gc.cpp
> > Lines 206 (patched)
> > <https://reviews.apache.org/r/53479/diff/6/?file=1776251#file1776251line214>
> >
> >     Kill this since we have the CHECK_READY.

Removed.


> On July 14, 2017, 8:47 a.m., Jiang Yan Xu wrote:
> > src/slave/gc.cpp
> > Lines 210 (patched)
> > <https://reviews.apache.org/r/53479/diff/6/?file=1776251#file1776251line218>
> >
> >     Use `CHECK_EQ(timeouts.erase(info->path), 1);`?

Also fixed this in `unschedule`


- Jacob


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/53479/#review180510
-----------------------------------------------------------


On July 15, 2017, 1:20 a.m., Jacob Janco wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/53479/
> -----------------------------------------------------------
> 
> (Updated July 15, 2017, 1:20 a.m.)
> 
> 
> Review request for mesos, Benjamin Mahler and Jiang Yan Xu.
> 
> 
> Bugs: MESOS-6549
>     https://issues.apache.org/jira/browse/MESOS-6549
> 
> 
> Repository: mesos
> 
> 
> Description
> -------
> 
> - Previously, rmdir operations were serialized
>   through the garbage collector. Expensive removal
>   events had the potential to delay task launch.
> - MESOS-6549
> 
> 
> Diffs
> -----
> 
>   src/slave/gc.hpp 5ea82456cffa374203371f227b6f0ee00296553e 
>   src/slave/gc.cpp 961f547793984449ea113d9664b12b5033638c58 
> 
> 
> Diff: https://reviews.apache.org/r/53479/diff/8/
> 
> 
> Testing
> -------
> 
> `./bin/mesos-tests.sh --gtest_filter="GarbageCollector*" --gtest_repeat=100 --gtest_break_on_failure`
> 
> 
> Thanks,
> 
> Jacob Janco
> 
>


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