mesos-reviews mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Benno Evers <bev...@mesosphere.com>
Subject Re: Review Request 64012: Added new --reconfiguration_compatibility slave flag and implementation.
Date Tue, 05 Dec 2017 15:44:49 GMT


> On Dec. 4, 2017, 11:14 p.m., Vinod Kone wrote:
> > src/tests/slave_tests.cpp
> > Lines 8912 (patched)
> > <https://reviews.apache.org/r/64012/diff/4-6/?file=1903901#file1903901line8920>
> >
> >     what about testing the resources?

I originally left that out because it has to assume quite a lot about the internal resource
handling of master and slave, which didn't seem very relevant to agent reconfiguration and
I thought it would be highly unlikely that a bug would be able to fail this and pass the test
in `master_tests.cpp` where we actually check the resources. Of course, it turned out that
there actually was a latent bug in the allocator which happens only in precisely the situation
we have here: https://reviews.apache.org/r/64342/


- Benno


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/64012/#review192788
-----------------------------------------------------------


On Dec. 4, 2017, 4:33 p.m., Benno Evers wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/64012/
> -----------------------------------------------------------
> 
> (Updated Dec. 4, 2017, 4:33 p.m.)
> 
> 
> Review request for mesos and Vinod Kone.
> 
> 
> Repository: mesos
> 
> 
> Description
> -------
> 
> This flag allows operators to weaken the checks performed by the agent
> when recovering state, in particular it allows to recover running tasks
> even when parts of the recovered SlaveInfo don't match the current
> state.
> 
> The set of possible changes is quite restricted for now,
> to avoid accidenetally introducing semantic bugs.
> 
> 
> Diffs
> -----
> 
>   src/CMakeLists.txt 592489df070a0c4fdee814a4a7b613e62a544f88 
>   src/Makefile.am 08d29ab58fe5c7f5e0c9173b504787ec254ed99b 
>   src/slave/compatibility.hpp PRE-CREATION 
>   src/slave/compatibility.cpp PRE-CREATION 
>   src/slave/flags.hpp f25d8aff29d0698eaf06dd40a86a35e43bc9e094 
>   src/slave/flags.cpp 34edf4cb4a1afc366e6b6b4ddf78e90e49853fdf 
>   src/slave/slave.hpp 643d8559244f0842e82b293c1ef99cb26e111059 
>   src/slave/slave.cpp c07e25f668bef6b6fa3cae4b69ba90de3eb3bdcc 
>   src/tests/CMakeLists.txt 5e5202079c9798b91028e845490b747d4d1d0a66 
>   src/tests/master_tests.cpp 08742ec878b29ef275fb40e857efffcf4c2f9ac2 
>   src/tests/slave_compatibility_tests.cpp PRE-CREATION 
>   src/tests/slave_recovery_tests.cpp 7674e607ee5bb7efbceb81c3e7291bc558d69935 
>   src/tests/slave_tests.cpp a7f6658fa467914cd74691b5631d26d85ba7c1eb 
> 
> 
> Diff: https://reviews.apache.org/r/64012/diff/6/
> 
> 
> Testing
> -------
> 
> 
> Thanks,
> 
> Benno Evers
> 
>


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