mesos-reviews mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Joseph Wu <jos...@mesosphere.io>
Subject Re: Review Request 60888: Added recovery logic for standalone containers.
Date Thu, 02 Nov 2017 15:49:24 GMT


> On Oct. 17, 2017, 10:39 a.m., Jie Yu wrote:
> > src/slave/containerizer/mesos/containerizer.cpp
> > Line 757 (original), 758 (patched)
> > <https://reviews.apache.org/r/60888/diff/3/?file=1858652#file1858652line758>
> >
> >     not yours, can you add a comment here saying that parent containerid will be
guaranteed to appear first in the vector. This is important for the following code to always
be able to find the parent container information for a nested container.

The ordering did not appear to matter for nested containers in the past, but now it does.
 The note I added explains why.


> On Oct. 17, 2017, 10:39 a.m., Jie Yu wrote:
> > src/slave/paths.hpp
> > Lines 53-54 (patched)
> > <https://reviews.apache.org/r/60888/diff/3/?file=1858655#file1858655line53>
> >
> >     Putting this path here is a bit wierd because this directory is kind of containerizer
specific.
> >     
> >     I'd suggest we create a `src/slave/containerizer/paths.hpp` to put those path
helper functions.
> >     
> >     `src/slave/paths.hpp` should probably only have paths that are related to slave,
but not containers. In the future, the sandbox for executors can be a symlink to the actual
container sandbox.

Leaving this open as I didn't change the directory structure...

Sandbox directories (for normal and standalone containers) are currently specified by the
agent, so it makes sense to have this directory in the agent's helper.  The containerizer
only specifies sandboxes for nested containers by appending onto an existing sandbox (which
itself is directly or indirectly (for multi-level nested containers) specified by the agent).


- Joseph


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/60888/#review188334
-----------------------------------------------------------


On Nov. 2, 2017, 8:43 a.m., Joseph Wu wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/60888/
> -----------------------------------------------------------
> 
> (Updated Nov. 2, 2017, 8:43 a.m.)
> 
> 
> Review request for mesos, Gilbert Song and Jie Yu.
> 
> 
> Bugs: MESOS-7305
>     https://issues.apache.org/jira/browse/MESOS-7305
> 
> 
> Repository: mesos
> 
> 
> Description
> -------
> 
> Although there is no way to launch standalone containers yet,
> this commit outlines the expected layout of container metadata
> which should be populated when launching standalone containers.
> 
> The layout is fairly simple, as standalone containers have no
> framework, executor, or tasks to worry about.  The sandbox directory
> will live under a new top-level directory `containers` and there is
> no metadata to checkpoint at the moment.
> 
> The containerizer will checkpoint a marker file (in the runtime
> directory) so that it knows to recover all standalone containers.
> 
> 
> Diffs
> -----
> 
>   src/slave/containerizer/mesos/containerizer.cpp 100e3bbda543d87808da9ff6bea42da5099ea8c5

>   src/slave/containerizer/mesos/paths.hpp a03f15e01e8eb8a1326baad8d7db96a79d785482 
>   src/slave/containerizer/mesos/paths.cpp 0c61c20c345a327ec469b382558aaeed0280e754 
>   src/slave/paths.hpp f000508d414daf9f943561f89c7105503a8a98b3 
>   src/slave/paths.cpp fd546525b900cb6524fb9196d19616ef18de0f30 
> 
> 
> Diff: https://reviews.apache.org/r/60888/diff/4/
> 
> 
> Testing
> -------
> 
> See next patch and later ones too.
> 
> 
> Thanks,
> 
> Joseph Wu
> 
>


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