mesos-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Ben Mahler" <benjamin.mah...@gmail.com>
Subject Re: Review Request 28720: Adjusted the calculation of unused resources in _launchTasks by considering persistent disk acquisition.
Date Sat, 13 Dec 2014 01:21:29 GMT

-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/28720/#review65007
-----------------------------------------------------------


I'd like to use the mutation stuff in the allocator, can you split it out? :D

Also, would love to see some tests for them once you pull them out!


src/master/master.cpp
<https://reviews.apache.org/r/28720/#comment107900>

    It's interesting to compare this to state::Variable's mutate() method, perhaps we should
call this transformation? Mutation sounds very general and could encapsulate things like addition
/ subtraction, which we already support.
    
    We'll need to expose this in order to allow the allocator to update resources as well.
Do you think this would benefit from being first class in resources.hpp?



src/master/master.cpp
<https://reviews.apache.org/r/28720/#comment107901>

    Maybe "sequence" is a better word than "series"?
    
    Do we need the reference to the composite pattern wikipedia page? Seems very straightforward
(and the wikipedia page is quite big) :)



src/master/master.cpp
<https://reviews.apache.org/r/28720/#comment107904>

    could we do s/_result/applied/ or s/_result/apply ?



src/master/master.cpp
<https://reviews.apache.org/r/28720/#comment107921>

    Do you need to be validating disk? Does it work if has_disk() == false?



src/master/master.cpp
<https://reviews.apache.org/r/28720/#comment107906>

    Hm.. we might want to document the "idempotent" nature of this operation at the top level
of this class.



src/master/master.cpp
<https://reviews.apache.org/r/28720/#comment107908>

    I think you mean non-persistent volumes are not supported..? And persistent volumes cannot
be acquired from existing non-persistent volumes, only from regular disk resources?



src/master/master.cpp
<https://reviews.apache.org/r/28720/#comment107916>

    Yikes! Could we match the wrong role here? Shouldn't we be looking for a role match?


- Ben Mahler


On Dec. 5, 2014, 6:01 p.m., Jie Yu wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/28720/
> -----------------------------------------------------------
> 
> (Updated Dec. 5, 2014, 6:01 p.m.)
> 
> 
> Review request for mesos and Ben Mahler.
> 
> 
> Bugs: MESOS-2030
>     https://issues.apache.org/jira/browse/MESOS-2030
> 
> 
> Repository: mesos-git
> 
> 
> Description
> -------
> 
> Introduced an abstraction for mutating resources. Acquiring persistent disk is one type
of resources mutation.
> 
> Infer persistent disk acquisitions from resources and check resource usage against adjusted
total resources.
> 
> Adjusted the calculation of unused resources in _launchTasks by considering persistent
disk acquisition.
> 
> 
> Diffs
> -----
> 
>   src/master/master.cpp 1cf2074b78e260bcccf96f4383bc4747b1e75063 
>   src/tests/resource_offers_tests.cpp e13b6c5460d9e6729843c40bed9e4d4e3f76d5d3 
> 
> Diff: https://reviews.apache.org/r/28720/diff/
> 
> 
> Testing
> -------
> 
> make check
> 
> 
> Thanks,
> 
> Jie Yu
> 
>


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