Re: [RFC] REST / thrift & AURORA-987

2015-12-01 Thread John Sirois 
On Tue, Dec 1, 2015 at 10:37 AM, David McLaughlin 
wrote:

> Shouldn't we start with the design of the API itself? Won't that influence
> many of the answers to these questions?
>
> E.g. if you're just looking to port the Thrift API 1:1 to a JSON + HTTP
> interface then that's a very different set of requirements to starting
> fresh and doing a better job with our API.
>
> Personally I don't think the existing Thrift API is a very good base to
> build an API on top off. A lot of the endpoints are fit for one purpose
> (e.g. a specific UI view or client function) rather than being flexible. I
> can't tell you how many times we wanted to go in and improve the UI in some
> way only to find the existing API does not give us access to the data we
> want.
>

Thanks - that is truly useful context;
https://issues.apache.org/jira/browse/AURORA-987 / the existing design doc
gave no real hint that there was pain with the _shape_ of the API.

Let me look further into shape issues - I haven't considered that at all
yet.  If you have any pointers or anecdotes about specific bits that caused
UI pain, those would speed me along.  I'm assuming the existing thrift api
then works fine for the python client, poorly for the UI.  If that's not
true and the api is also painful for the python client, that would be
useful to know as well.


> So yeah, I feel like the API should be more generic with regards to data
> access. So fewer, more-powerful endpoints that support complex queries.
>
> On Mon, Nov 30, 2015 at 12:16 PM, John Sirois 
> wrote:
>
> > I’ve experimenting on https://issues.apache.org/jira/browse/AURORA-987
> for
> > the past few weeks and I’d like to ask for feedback on the direction I’d
> > like to head. If you’re interested in the evolution of the Aurora REST
> api,
> > read on.
> > --
> >
> > AURORA-987 aims to create a first-class REST-like scheduler interface.
> I’ve
> > re-familiarized myself with the codebase and come to the conclusion that
> > transitioning to a 1st class REST api requires maintaining the core
> thrift
> > API as the 1st class API until the point the REST API is fully
> established
> > and clients can all be transitioned.
> >
> > I think this conclusion is probably uncontroversial, but the key factors
> > pushing this way are:
> >
> >1.
> >
> >The thrift API has both wide and deep dependencies inside the Aurora
> >codebase - 276 imports across 97 files:
> >
> >$ git grep "import org.apache.aurora.gen" -- src/main/java/ | grep
> > -v "import org.apache.aurora.gen.storage" | wc -l
> >276
> >$ git grep "import org.apache.aurora.gen" -- src/main/java/ | grep
> > -v "import org.apache.aurora.gen.storage" | cut -d: -f1 | sort -u | wc
> > -l
> >97
> >
> >2.
> >
> >The thrift API is stored long-term in the log in serialized form.
> >
> > Both 1 & 2 dictate that the thrift API, at least its enums, structs and
> > unions, must be maintained for the forseeable future.
> > We also have the RPC API (thrift services) - which is currently a ~thin,
> > but not insignificant, container of API processing logic. For example,
> see
> > here
> > <
> >
> https://github.com/apache/aurora/blob/master/src/main/java/org/apache/aurora/scheduler/thrift/SchedulerThriftInterface.java#L220-L267
> > >
> > .
> >
> > As such it seems to me the REST API should call into the existing thrift
> > API to provide a stable transition and confidence in core logic of API
> > method implementations.
> >
> > This leads to the following ideas for paths forward:
> >
> >1. Hand construct a REST forwarding layer and maintain it in tandem
> with
> >thrift API changes.
> >2. Automate 1 such that thrift API changes cause REST API changes
> >automatically.
> >
> > The hand construction path has the obvious maintenance issues, but is
> > otherwise straight-forward. The maintenance issues should not be
> > overstated, since good tests and some extra review vigilance could be
> > enough to make the approach work for the period of time both APIs are
> > supported.
> >
> > That said, an automated solution with a single source of truth for the
> API
> > definition is clearly preferrable given the automation is free.
> > The automation is far from free though and so I’ve started investigating
> > one approach to this automation to flesh out the scope.
> >
> > We already do our own thrift codegen
> > <
> >
> https://github.com/apache/aurora/blob/master/src/main/python/apache/aurora/tools/java/thrift_wrapper_codegen.py
> > >
> > via a custom gradle ThriftEntitiesPlugin
> > <
> >
> https://github.com/apache/aurora/blob/master/buildSrc/src/main/groovy/org/apache/aurora/build/ThriftEntitiesPlugin.groovy
> > >
> > that works around Apache thrift’s java codegen in order to generate
> > immutable wrapper entities for the storage system.
> > I propose taking this further and generating our own thrift API and
> > entities in 1 pass

Re: [RFC] REST / thrift & AURORA-987

2015-12-01 Thread John Sirois 
On Tue, Dec 1, 2015 at 11:21 AM, John Sirois  wrote:

>
>
> On Tue, Dec 1, 2015 at 11:17 AM, Igor Morozov  wrote:
>
>> We had very similar concerns here at Uber in regards of Thrift and
>> newer REST API that is coming to Aurora scheduler. It feels like
>> having a general API model in a scheduler and providing whatever
>> interface is necessary/convenient for integration would generally be a
>> better option then building REST API layer on top of Thrift API.
>>
>
> To be clear, my intent was to build on top in phase 1, then back out the
> thrift API and gut it as phase 2, then evolve in phase 3.
>
> That said, its clear from both your comment and David's that there is a
> desire to go straight to a new API side-by side with the existing API 1st,
> then transition clients, then gut thrift.
>

... I guess those phasings are similar- the key difference is when the new
API is published / blessed.  My original plan was to do this at the end of
phase 3, you two are suggesting do this at the end of phase 1.


> Igor, if you also have any specifics on problematic bits of the current
> API - I'd love to have those.
>
>
>>
>> On Tue, Dec 1, 2015 at 9:37 AM, David McLaughlin 
>> wrote:
>> > Shouldn't we start with the design of the API itself? Won't that
>> influence
>> > many of the answers to these questions?
>> >
>> > E.g. if you're just looking to port the Thrift API 1:1 to a JSON + HTTP
>> > interface then that's a very different set of requirements to starting
>> > fresh and doing a better job with our API.
>> >
>> > Personally I don't think the existing Thrift API is a very good base to
>> > build an API on top off. A lot of the endpoints are fit for one purpose
>> > (e.g. a specific UI view or client function) rather than being
>> flexible. I
>> > can't tell you how many times we wanted to go in and improve the UI in
>> some
>> > way only to find the existing API does not give us access to the data we
>> > want.
>> >
>> > So yeah, I feel like the API should be more generic with regards to data
>> > access. So fewer, more-powerful endpoints that support complex queries.
>> >
>> > On Mon, Nov 30, 2015 at 12:16 PM, John Sirois 
>> wrote:
>> >
>> >> I’ve experimenting on https://issues.apache.org/jira/browse/AURORA-987
>> for
>> >> the past few weeks and I’d like to ask for feedback on the direction
>> I’d
>> >> like to head. If you’re interested in the evolution of the Aurora REST
>> api,
>> >> read on.
>> >> --
>> >>
>> >> AURORA-987 aims to create a first-class REST-like scheduler interface.
>> I’ve
>> >> re-familiarized myself with the codebase and come to the conclusion
>> that
>> >> transitioning to a 1st class REST api requires maintaining the core
>> thrift
>> >> API as the 1st class API until the point the REST API is fully
>> established
>> >> and clients can all be transitioned.
>> >>
>> >> I think this conclusion is probably uncontroversial, but the key
>> factors
>> >> pushing this way are:
>> >>
>> >>1.
>> >>
>> >>The thrift API has both wide and deep dependencies inside the Aurora
>> >>codebase - 276 imports across 97 files:
>> >>
>> >>$ git grep "import org.apache.aurora.gen" -- src/main/java/ | grep
>> >> -v "import org.apache.aurora.gen.storage" | wc -l
>> >>276
>> >>$ git grep "import org.apache.aurora.gen" -- src/main/java/ | grep
>> >> -v "import org.apache.aurora.gen.storage" | cut -d: -f1 | sort -u | wc
>> >> -l
>> >>97
>> >>
>> >>2.
>> >>
>> >>The thrift API is stored long-term in the log in serialized form.
>> >>
>> >> Both 1 & 2 dictate that the thrift API, at least its enums, structs and
>> >> unions, must be maintained for the forseeable future.
>> >> We also have the RPC API (thrift services) - which is currently a
>> ~thin,
>> >> but not insignificant, container of API processing logic. For example,
>> see
>> >> here
>> >> <
>> >>
>> https://github.com/apache/aurora/blob/master/src/main/java/org/apache/aurora/scheduler/thrift/SchedulerThriftInterface.java#L220-L267
>> >> >
>> >> .
>> >>
>> >> As such it seems to me the REST API should call into the existing
>> thrift
>> >> API to provide a stable transition and confidence in core logic of API
>> >> method implementations.
>> >>
>> >> This leads to the following ideas for paths forward:
>> >>
>> >>1. Hand construct a REST forwarding layer and maintain it in tandem
>> with
>> >>thrift API changes.
>> >>2. Automate 1 such that thrift API changes cause REST API changes
>> >>automatically.
>> >>
>> >> The hand construction path has the obvious maintenance issues, but is
>> >> otherwise straight-forward. The maintenance issues should not be
>> >> overstated, since good tests and some extra review vigilance could be
>> >> enough to make the approach work for the period of time both APIs are
>> >> supported.
>> >>
>> >> That said, an automated solution with a single source of

Re: [RFC] REST / thrift & AURORA-987

2015-12-01 Thread John Sirois 
On Tue, Dec 1, 2015 at 11:17 AM, Igor Morozov  wrote:

> We had very similar concerns here at Uber in regards of Thrift and
> newer REST API that is coming to Aurora scheduler. It feels like
> having a general API model in a scheduler and providing whatever
> interface is necessary/convenient for integration would generally be a
> better option then building REST API layer on top of Thrift API.
>

To be clear, my intent was to build on top in phase 1, then back out the
thrift API and gut it as phase 2, then evolve in phase 3.

That said, its clear from both your comment and David's that there is a
desire to go straight to a new API side-by side with the existing API 1st,
then transition clients, then gut thrift.

Igor, if you also have any specifics on problematic bits of the current API
- I'd love to have those.


>
> On Tue, Dec 1, 2015 at 9:37 AM, David McLaughlin 
> wrote:
> > Shouldn't we start with the design of the API itself? Won't that
> influence
> > many of the answers to these questions?
> >
> > E.g. if you're just looking to port the Thrift API 1:1 to a JSON + HTTP
> > interface then that's a very different set of requirements to starting
> > fresh and doing a better job with our API.
> >
> > Personally I don't think the existing Thrift API is a very good base to
> > build an API on top off. A lot of the endpoints are fit for one purpose
> > (e.g. a specific UI view or client function) rather than being flexible.
> I
> > can't tell you how many times we wanted to go in and improve the UI in
> some
> > way only to find the existing API does not give us access to the data we
> > want.
> >
> > So yeah, I feel like the API should be more generic with regards to data
> > access. So fewer, more-powerful endpoints that support complex queries.
> >
> > On Mon, Nov 30, 2015 at 12:16 PM, John Sirois 
> wrote:
> >
> >> I’ve experimenting on https://issues.apache.org/jira/browse/AURORA-987
> for
> >> the past few weeks and I’d like to ask for feedback on the direction I’d
> >> like to head. If you’re interested in the evolution of the Aurora REST
> api,
> >> read on.
> >> --
> >>
> >> AURORA-987 aims to create a first-class REST-like scheduler interface.
> I’ve
> >> re-familiarized myself with the codebase and come to the conclusion that
> >> transitioning to a 1st class REST api requires maintaining the core
> thrift
> >> API as the 1st class API until the point the REST API is fully
> established
> >> and clients can all be transitioned.
> >>
> >> I think this conclusion is probably uncontroversial, but the key factors
> >> pushing this way are:
> >>
> >>1.
> >>
> >>The thrift API has both wide and deep dependencies inside the Aurora
> >>codebase - 276 imports across 97 files:
> >>
> >>$ git grep "import org.apache.aurora.gen" -- src/main/java/ | grep
> >> -v "import org.apache.aurora.gen.storage" | wc -l
> >>276
> >>$ git grep "import org.apache.aurora.gen" -- src/main/java/ | grep
> >> -v "import org.apache.aurora.gen.storage" | cut -d: -f1 | sort -u | wc
> >> -l
> >>97
> >>
> >>2.
> >>
> >>The thrift API is stored long-term in the log in serialized form.
> >>
> >> Both 1 & 2 dictate that the thrift API, at least its enums, structs and
> >> unions, must be maintained for the forseeable future.
> >> We also have the RPC API (thrift services) - which is currently a ~thin,
> >> but not insignificant, container of API processing logic. For example,
> see
> >> here
> >> <
> >>
> https://github.com/apache/aurora/blob/master/src/main/java/org/apache/aurora/scheduler/thrift/SchedulerThriftInterface.java#L220-L267
> >> >
> >> .
> >>
> >> As such it seems to me the REST API should call into the existing thrift
> >> API to provide a stable transition and confidence in core logic of API
> >> method implementations.
> >>
> >> This leads to the following ideas for paths forward:
> >>
> >>1. Hand construct a REST forwarding layer and maintain it in tandem
> with
> >>thrift API changes.
> >>2. Automate 1 such that thrift API changes cause REST API changes
> >>automatically.
> >>
> >> The hand construction path has the obvious maintenance issues, but is
> >> otherwise straight-forward. The maintenance issues should not be
> >> overstated, since good tests and some extra review vigilance could be
> >> enough to make the approach work for the period of time both APIs are
> >> supported.
> >>
> >> That said, an automated solution with a single source of truth for the
> API
> >> definition is clearly preferrable given the automation is free.
> >> The automation is far from free though and so I’ve started investigating
> >> one approach to this automation to flesh out the scope.
> >>
> >> We already do our own thrift codegen
> >> <
> >>
> https://github.com/apache/aurora/blob/master/src/main/python/apache/aurora/tools/java/thrift_wrapper_codegen.py
> >> >
> >> via a custom gradle

Re: [RFC] REST / thrift & AURORA-987

2015-12-01 Thread Joshua Cohen 
John: I share David's concerns, but it's not clear to me that they're
incompatible with your proposal. I.e. we could design the new REST API
according to any interface we like and still implement that API on top of
the thrift API (though the desired interface should always be our guide, we
should not be constrained by what's currently possible to build on top of
the thrift API). That said, I think it would be beneficial to start with
the API design and work from there back to the implementation.

On Tue, Dec 1, 2015 at 12:31 PM, Igor Morozov  wrote:

> We'll be sharing all the questions in single concise manner after we
> get the result of out POC(we're currently integrating aurora scheduler
> with internal deployment system so more questions are coming ;) )
>
> Thanks for asking!
>
> -Igor
>
> On Tue, Dec 1, 2015 at 10:21 AM, John Sirois  wrote:
> > On Tue, Dec 1, 2015 at 11:17 AM, Igor Morozov  wrote:
> >
> >> We had very similar concerns here at Uber in regards of Thrift and
> >> newer REST API that is coming to Aurora scheduler. It feels like
> >> having a general API model in a scheduler and providing whatever
> >> interface is necessary/convenient for integration would generally be a
> >> better option then building REST API layer on top of Thrift API.
> >>
> >
> > To be clear, my intent was to build on top in phase 1, then back out the
> > thrift API and gut it as phase 2, then evolve in phase 3.
> >
> > That said, its clear from both your comment and David's that there is a
> > desire to go straight to a new API side-by side with the existing API
> 1st,
> > then transition clients, then gut thrift.
> >
> > Igor, if you also have any specifics on problematic bits of the current
> API
> > - I'd love to have those.
> >
> >
> >>
> >> On Tue, Dec 1, 2015 at 9:37 AM, David McLaughlin <
> dmclaugh...@apache.org>
> >> wrote:
> >> > Shouldn't we start with the design of the API itself? Won't that
> >> influence
> >> > many of the answers to these questions?
> >> >
> >> > E.g. if you're just looking to port the Thrift API 1:1 to a JSON +
> HTTP
> >> > interface then that's a very different set of requirements to starting
> >> > fresh and doing a better job with our API.
> >> >
> >> > Personally I don't think the existing Thrift API is a very good base
> to
> >> > build an API on top off. A lot of the endpoints are fit for one
> purpose
> >> > (e.g. a specific UI view or client function) rather than being
> flexible.
> >> I
> >> > can't tell you how many times we wanted to go in and improve the UI in
> >> some
> >> > way only to find the existing API does not give us access to the data
> we
> >> > want.
> >> >
> >> > So yeah, I feel like the API should be more generic with regards to
> data
> >> > access. So fewer, more-powerful endpoints that support complex
> queries.
> >> >
> >> > On Mon, Nov 30, 2015 at 12:16 PM, John Sirois 
> >> wrote:
> >> >
> >> >> I’ve experimenting on
> https://issues.apache.org/jira/browse/AURORA-987
> >> for
> >> >> the past few weeks and I’d like to ask for feedback on the direction
> I’d
> >> >> like to head. If you’re interested in the evolution of the Aurora
> REST
> >> api,
> >> >> read on.
> >> >> --
> >> >>
> >> >> AURORA-987 aims to create a first-class REST-like scheduler
> interface.
> >> I’ve
> >> >> re-familiarized myself with the codebase and come to the conclusion
> that
> >> >> transitioning to a 1st class REST api requires maintaining the core
> >> thrift
> >> >> API as the 1st class API until the point the REST API is fully
> >> established
> >> >> and clients can all be transitioned.
> >> >>
> >> >> I think this conclusion is probably uncontroversial, but the key
> factors
> >> >> pushing this way are:
> >> >>
> >> >>1.
> >> >>
> >> >>The thrift API has both wide and deep dependencies inside the
> Aurora
> >> >>codebase - 276 imports across 97 files:
> >> >>
> >> >>$ git grep "import org.apache.aurora.gen" -- src/main/java/ | grep
> >> >> -v "import org.apache.aurora.gen.storage" | wc -l
> >> >>276
> >> >>$ git grep "import org.apache.aurora.gen" -- src/main/java/ | grep
> >> >> -v "import org.apache.aurora.gen.storage" | cut -d: -f1 | sort -u |
> wc
> >> >> -l
> >> >>97
> >> >>
> >> >>2.
> >> >>
> >> >>The thrift API is stored long-term in the log in serialized form.
> >> >>
> >> >> Both 1 & 2 dictate that the thrift API, at least its enums, structs
> and
> >> >> unions, must be maintained for the forseeable future.
> >> >> We also have the RPC API (thrift services) - which is currently a
> ~thin,
> >> >> but not insignificant, container of API processing logic. For
> example,
> >> see
> >> >> here
> >> >> <
> >> >>
> >>
> https://github.com/apache/aurora/blob/master/src/main/java/org/apache/aurora/scheduler/thrift/SchedulerThriftInterface.java#L220-L267
> >> >> >
> >> >> .
> >> >>
> >> >> As such it seems to me the

Re: [RFC] REST / thrift & AURORA-987

2015-12-01 Thread John Sirois 
On Tue, Dec 1, 2015 at 11:55 AM, Joshua Cohen  wrote:

> John: I share David's concerns, but it's not clear to me that they're
> incompatible with your proposal. I.e. we could design the new REST API
> according to any interface we like and still implement that API on top of
> the thrift API (though the desired interface should always be our guide, we
> should not be constrained by what's currently possible to build on top of
> the thrift API). That said, I think it would be beneficial to start with
> the API design and work from there back to the implementation.
>

Sounds great.  Totally on board with this, I just lack context at the
moment and will ramp up starting today on API shape.
I'm your context-free guy to solve problems in the codebase, but I'll be a
slower helper, more of a ramp-up burden on this sort of thing that requires
gathering the shared pain / hopes / dreams knowledge you all have lived
through.


> On Tue, Dec 1, 2015 at 12:31 PM, Igor Morozov  wrote:
>
> > We'll be sharing all the questions in single concise manner after we
> > get the result of out POC(we're currently integrating aurora scheduler
> > with internal deployment system so more questions are coming ;) )
> >
> > Thanks for asking!
> >
> > -Igor
> >
> > On Tue, Dec 1, 2015 at 10:21 AM, John Sirois 
> wrote:
> > > On Tue, Dec 1, 2015 at 11:17 AM, Igor Morozov 
> wrote:
> > >
> > >> We had very similar concerns here at Uber in regards of Thrift and
> > >> newer REST API that is coming to Aurora scheduler. It feels like
> > >> having a general API model in a scheduler and providing whatever
> > >> interface is necessary/convenient for integration would generally be a
> > >> better option then building REST API layer on top of Thrift API.
> > >>
> > >
> > > To be clear, my intent was to build on top in phase 1, then back out
> the
> > > thrift API and gut it as phase 2, then evolve in phase 3.
> > >
> > > That said, its clear from both your comment and David's that there is a
> > > desire to go straight to a new API side-by side with the existing API
> > 1st,
> > > then transition clients, then gut thrift.
> > >
> > > Igor, if you also have any specifics on problematic bits of the current
> > API
> > > - I'd love to have those.
> > >
> > >
> > >>
> > >> On Tue, Dec 1, 2015 at 9:37 AM, David McLaughlin <
> > dmclaugh...@apache.org>
> > >> wrote:
> > >> > Shouldn't we start with the design of the API itself? Won't that
> > >> influence
> > >> > many of the answers to these questions?
> > >> >
> > >> > E.g. if you're just looking to port the Thrift API 1:1 to a JSON +
> > HTTP
> > >> > interface then that's a very different set of requirements to
> starting
> > >> > fresh and doing a better job with our API.
> > >> >
> > >> > Personally I don't think the existing Thrift API is a very good base
> > to
> > >> > build an API on top off. A lot of the endpoints are fit for one
> > purpose
> > >> > (e.g. a specific UI view or client function) rather than being
> > flexible.
> > >> I
> > >> > can't tell you how many times we wanted to go in and improve the UI
> in
> > >> some
> > >> > way only to find the existing API does not give us access to the
> data
> > we
> > >> > want.
> > >> >
> > >> > So yeah, I feel like the API should be more generic with regards to
> > data
> > >> > access. So fewer, more-powerful endpoints that support complex
> > queries.
> > >> >
> > >> > On Mon, Nov 30, 2015 at 12:16 PM, John Sirois <
> john.sir...@gmail.com>
> > >> wrote:
> > >> >
> > >> >> I’ve experimenting on
> > https://issues.apache.org/jira/browse/AURORA-987
> > >> for
> > >> >> the past few weeks and I’d like to ask for feedback on the
> direction
> > I’d
> > >> >> like to head. If you’re interested in the evolution of the Aurora
> > REST
> > >> api,
> > >> >> read on.
> > >> >> --
> > >> >>
> > >> >> AURORA-987 aims to create a first-class REST-like scheduler
> > interface.
> > >> I’ve
> > >> >> re-familiarized myself with the codebase and come to the conclusion
> > that
> > >> >> transitioning to a 1st class REST api requires maintaining the core
> > >> thrift
> > >> >> API as the 1st class API until the point the REST API is fully
> > >> established
> > >> >> and clients can all be transitioned.
> > >> >>
> > >> >> I think this conclusion is probably uncontroversial, but the key
> > factors
> > >> >> pushing this way are:
> > >> >>
> > >> >>1.
> > >> >>
> > >> >>The thrift API has both wide and deep dependencies inside the
> > Aurora
> > >> >>codebase - 276 imports across 97 files:
> > >> >>
> > >> >>$ git grep "import org.apache.aurora.gen" -- src/main/java/ |
> grep
> > >> >> -v "import org.apache.aurora.gen.storage" | wc -l
> > >> >>276
> > >> >>$ git grep "import org.apache.aurora.gen" -- src/main/java/ |
> grep
> > >> >> -v "import org.apache.aurora.gen.storage" | cut -d: -f1 | sort -u |
> > wc
> > >> >> -l
> >