> On July 6, 2015, 6:31 p.m., Kevin Sweeney wrote:
> > +1 to this tactical fix. Which APIs need to be changed to use slaveID 
> > instead of hostName as the canonical identifier for mesos slaves within 
> > aurora?

I'd approach bottom-up, starting with `AttributeStore#getHostAttributes()`.  
Most callers seem like they could pretty easily use slave ID instead of host 
name for this.  It only gets tricky with maintenance-related callers; we'd just 
need to figure out how to behave when multiple slaves are on a host.  I'll add 
a TODO for this.


- Bill


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/36096/#review90523
-----------------------------------------------------------


On July 1, 2015, 8:01 p.m., Bill Farner wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/36096/
> -----------------------------------------------------------
> 
> (Updated July 1, 2015, 8:01 p.m.)
> 
> 
> Review request for Aurora and Kevin Sweeney.
> 
> 
> Bugs: AURORA-1379
>     https://issues.apache.org/jira/browse/AURORA-1379
> 
> 
> Repository: aurora
> 
> 
> Description
> -------
> 
> The idea of this change is to avoid displacing the top-level record in the 
> `host_attributes` table when mutating, but instead mutate specific fields.
> 
> 
> Diffs
> -----
> 
>   src/main/java/org/apache/aurora/scheduler/storage/db/AttributeMapper.java 
> 3763f4dbd1b83b83c4b19a89784343ef9de100f1 
>   src/main/java/org/apache/aurora/scheduler/storage/db/DbAttributeStore.java 
> 5f1cd2b7daec40a8a57095e16390698dbb0bcd24 
>   
> src/main/resources/org/apache/aurora/scheduler/storage/db/AttributeMapper.xml 
> d36b42a22ebf9bbe52e60fea9b8a2677c41bd598 
>   
> src/test/java/org/apache/aurora/scheduler/storage/db/DbAttributeStoreTest.java
>  db508cace922b20a19081a1d918cc082b255cd02 
> 
> Diff: https://reviews.apache.org/r/36096/diff/
> 
> 
> Testing
> -------
> 
> 
> Thanks,
> 
> Bill Farner
> 
>

Reply via email to