> On June 23, 2014, 8:34 p.m., Bill Farner wrote:
> > docs/sla.md, line 1
> > <https://reviews.apache.org/r/22697/diff/1/?file=611838#file611838line1>
> >
> >     How about "Aurora SLA Measurement"?  Given that this works above a job, 
> > broadened scope seems like a good idea.

Sure.


> On June 23, 2014, 8:34 p.m., Bill Farner wrote:
> > docs/sla.md, line 23
> > <https://reviews.apache.org/r/22697/diff/1/?file=611838#file611838line23>
> >
> >     s/core //?

Done.


> On June 23, 2014, 8:34 p.m., Bill Farner wrote:
> > docs/sla.md, line 21
> > <https://reviews.apache.org/r/22697/diff/1/?file=611838#file611838line21>
> >
> >     Might want to eschew discussion of implementation details a bit.  How 
> > about "Counters that track SLA measurements are computed periodically 
> > within the scheduler."

Done.


> On June 23, 2014, 8:34 p.m., Bill Farner wrote:
> > docs/sla.md, line 31
> > <https://reviews.apache.org/r/22697/diff/1/?file=611838#file611838line31>
> >
> >     Before this, a bit of prose about how we detect platform unavailability 
> > would be useful.  i.e. it's a red flag that the term LOST does not appear 
> > in the document.  It's good that you link to the source of truth 
> > (SlaAlgorithm.java), but a sentence or two about mesos and LOST status 
> > would fill in a big contextual blank.

Extended the definition and added a few examples of downtime events below.


> On June 23, 2014, 8:34 p.m., Bill Farner wrote:
> > docs/sla.md, line 45
> > <https://reviews.apache.org/r/22697/diff/1/?file=611838#file611838line45>
> >
> >     Nit: a user can indirectly cause service downtime when their 
> > application exits.

I think that would be considered as "unhealthy" service task. Modified.


> On June 23, 2014, 8:34 p.m., Bill Farner wrote:
> > docs/sla.md, line 51
> > <https://reviews.apache.org/r/22697/diff/1/?file=611838#file611838line51>
> >
> >     You can actually link directly to the source path for simpler discovery.

Done.


> On June 23, 2014, 8:34 p.m., Bill Farner wrote:
> > docs/sla.md, line 88
> > <https://reviews.apache.org/r/22697/diff/1/?file=611838#file611838line88>
> >
> >     s/Average/Median/
> >     
> >     s/get host-assigned/be assigned to a host/
> >     s/A combined/This is a combined/

Done.


> On June 23, 2014, 8:34 p.m., Bill Farner wrote:
> > docs/sla.md, line 89
> > <https://reviews.apache.org/r/22697/diff/1/?file=611838#file611838line89>
> >
> >     Rewording:
> >     "job scheduling performance dependency"
> >     
> >     to
> >     
> >     "the dependency of scheduling performance"

Done.


> On June 23, 2014, 8:34 p.m., Bill Farner wrote:
> > docs/sla.md, line 96
> > <https://reviews.apache.org/r/22697/diff/1/?file=611838#file611838line96>
> >
> >     Please link to the source defining these sizes.

Done.


> On June 23, 2014, 8:34 p.m., Bill Farner wrote:
> > docs/sla.md, line 118
> > <https://reviews.apache.org/r/22697/diff/1/?file=611838#file611838line118>
> >
> >     s/The //

Done.


> On June 23, 2014, 8:34 p.m., Bill Farner wrote:
> > docs/sla.md, line 124
> > <https://reviews.apache.org/r/22697/diff/1/?file=611838#file611838line124>
> >
> >     s/Average/Median/

Done.


- Maxim


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/22697/#review46442
-----------------------------------------------------------


On June 17, 2014, 9:14 p.m., Maxim Khutornenko wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/22697/
> -----------------------------------------------------------
> 
> (Updated June 17, 2014, 9:14 p.m.)
> 
> 
> Review request for Aurora.
> 
> 
> Bugs: AURORA-528
>     https://issues.apache.org/jira/browse/AURORA-528
> 
> 
> Repository: aurora
> 
> 
> Description
> -------
> 
> Documenting SLA stats.
> 
> 
> Diffs
> -----
> 
>   docs/sla.md PRE-CREATION 
> 
> Diff: https://reviews.apache.org/r/22697/diff/
> 
> 
> Testing
> -------
> 
> 
> Thanks,
> 
> Maxim Khutornenko
> 
>

Reply via email to