mesos-reviews mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Kevin Klues <klue...@gmail.com>
Subject Re: Review Request 51407: Updated mesos containerizer to checkpoint container runtime information.
Date Sat, 24 Sep 2016 20:01:35 GMT


> On Sept. 24, 2016, 1:50 a.m., Jie Yu wrote:
> > src/slave/containerizer/mesos/containerizer.cpp, lines 1253-1270
> > <https://reviews.apache.org/r/51407/diff/7/?file=1509433#file1509433line1253>
> >
> >     I think we should do that before provisioner/isolators prepare are called. The
reason is because we need to do cleanups for provisioner/isolators to undo what's done in
prepare after agent restarts if agent crashes after prepare is done but before here. This
is not important for top level containers because agent will checkpoint it anyway. But this
is important for nested containers, because agent won't checkpoint them.

The only advantage I see in moving it up is that we can put an optimization in the recover
path to treat any containers that actually have this directory created for them as "known
orphans". This will allow their destruction to be deferred to the bottom of the containerizer.recover()
path. Otherwise, they will be treated as "unknown orphans" and the provisioner/isolators will
have to do cleanup on them inline during their own recover() calls. Seems like a reasonable
optimization. I'l make the change.


- Kevin


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/51407/#review150282
-----------------------------------------------------------


On Sept. 23, 2016, 9:01 p.m., Kevin Klues wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/51407/
> -----------------------------------------------------------
> 
> (Updated Sept. 23, 2016, 9:01 p.m.)
> 
> 
> Review request for mesos and Jie Yu.
> 
> 
> Bugs: MESOS-6204
>     https://issues.apache.org/jira/browse/MESOS-6204
> 
> 
> Repository: mesos
> 
> 
> Description
> -------
> 
> This includes checkpointing both the container pid and the status of
> the container upon exit. This also includes an update to tests to
> account for new 'init' process semantics in a container. That is, the
> name of the init process of the container is now "mesos-containerizer"
> not "sh".
> 
> 
> Diffs
> -----
> 
>   src/slave/containerizer/mesos/containerizer.hpp 16f9e3e92e90fe7f8a0ebd24e567800e1f285bc9

>   src/slave/containerizer/mesos/containerizer.cpp 144b0db501d40d4e0bba12672723616bedd76e7e

>   src/tests/containerizer/isolator_tests.cpp b4d25e57df7f0e157769c9ae4f7847657c505e78

> 
> Diff: https://reviews.apache.org/r/51407/diff/
> 
> 
> Testing
> -------
> 
> $ GTEST_FILTER="" make -j check
> $ src/mesos-tests
> $ sudo src/mesos-tests
> 
> 
> Thanks,
> 
> Kevin Klues
> 
>


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