> On Aug. 1, 2014, 10:02 p.m., Bill Farner wrote: > > src/main/thrift/org/apache/aurora/gen/api.thrift, line 541 > > <https://reviews.apache.org/r/24116/diff/5/?file=648844#file648844line541> > > > > I wonder if we should use this opportunity to choose a name that does > > not imply batching. maxSimultaneousInstances? updateGroupSize? I'm hope > > to ideas. Anybody else have an opinion? We don't need to dwell on this. > > David McLaughlin wrote: > +1 for a name that more accurately reflects what the property does. > > Maxim Khutornenko wrote: > I am open to suggestions. Feels like updateGroupSize would be the least > tied to implementation.
+1 to updateGroupSize > On Aug. 1, 2014, 10:02 p.m., Bill Farner wrote: > > src/main/thrift/org/apache/aurora/gen/api.thrift, line 543 > > <https://reviews.apache.org/r/24116/diff/5/?file=648844#file648844line543> > > > > maxPerInstanceFailures? "total" to me suggests the "global counter", > > which i think we should avoid > > Maxim Khutornenko wrote: > This is the global counter. There must be something to track failed > instances. How about maxFailedInstances? > > Also, changed the maxInstanceFailures to maxPerInstanceFailures to avoid > confusion. I think we mean the same thing. maxFailedInstances and maxPerInstanceFailures should self-doc > On Aug. 1, 2014, 10:02 p.m., Bill Farner wrote: > > src/main/thrift/org/apache/aurora/gen/api.thrift, line 566 > > <https://reviews.apache.org/r/24116/diff/5/?file=648844#file648844line566> > > > > how would you feel about a closed range instead of a set? i think this > > improves API ergonomics, matches the use cases we know of, and doesn't > > break any use case. > > Maxim Khutornenko wrote: > The closed range would not address patchy cases like 2-7, 10-12, 15-18. > Also, it will implicitly assign TaskConfigs to instance "holes". > ...but you have a list of them a level above, so you would have list<range> instead of list<set>, both can express the same things AFAICT - Bill ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/24116/#review49387 ----------------------------------------------------------- On Aug. 1, 2014, 10:53 p.m., Maxim Khutornenko wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/24116/ > ----------------------------------------------------------- > > (Updated Aug. 1, 2014, 10:53 p.m.) > > > Review request for Aurora, David McLaughlin, Kevin Sweeney, and Bill Farner. > > > Bugs: AURORA-611 > https://issues.apache.org/jira/browse/AURORA-611 > > > Repository: aurora > > > Description > ------- > > First stab at update APIs. > > > Diffs > ----- > > > src/main/java/org/apache/aurora/scheduler/thrift/SchedulerThriftInterface.java > 12de5a3e9e3aae217b30c385e2d7ec7b43863ae2 > src/main/thrift/org/apache/aurora/gen/api.thrift > 54b8985971719247a5d42d8676075a51045bbb92 > src/test/java/org/apache/aurora/scheduler/thrift/aop/ForwardingThrift.java > 2ea4a9ba0a1ea81fea5c4f5203457aa79ae67c10 > > Diff: https://reviews.apache.org/r/24116/diff/ > > > Testing > ------- > > gradle build > > > Thanks, > > Maxim Khutornenko > >