[openstack-dev] [Nova] no-db-compute, a new service

Russell Bryant rbryant at redhat.com
Mon Nov 12 13:57:08 UTC 2012


On 11/12/2012 05:54 AM, Thierry Carrez wrote:
> Russell Bryant wrote:
>> This service would act as a proxy for nova-compute in a couple of ways.
>>
>> 1) The nova-compute service would use it as a proxy to accomplish
>> certain tasks, such as targeted operations that need database access.
>>
>> 2) Over time, it would be used as a proxy for other services that wish
>> to execute some sort of compute action.  For example, instead of
>> directly asking nova-compute to complete a long running operation, the
>> nova-compute-proxy would take this operation and monitor its progress.
> 
> I'm not convinced nova-compute-proxy is the best name for this. It's a
> bit of a mouthful and IMHO proxying has a "one-way" meaning into it...
> If the proxying was only in one direction (compute -> db) that would
> make perfect sense but if you add other more complex abstractions (like
> nodes -> compute), I think "proxy" is a bit reductive. nova-controller ?

nova-controller is fine with me ... naming is hard.

> Otherwise my main gripe with it would be that this new service would be
> compute-node-oriented, whereas the "nova-sink" approach was a bit more
> generic (could one day be called by other nodes to access DB). But I
> suspect you had the choice between the two approaches covered in the
> summit session I missed :)
> 

I suppose if it wasn't called "nova-compute-proxy" it's not artificially
limited to compute.  It could still be used for those other things.

The proposal in the summit session was nova-sink.  The feedback wasn't
that it was bad, but just that it didn't do enough.  Folks wanted a path
that would allow us to continue to do some more drastic work to
nova-compute, such as orchestrating long running tasks from a single
place so we can more easily ensure forward progress.

What I'm proposing now is something not called "nova-sink" but looks
just like it in the short term, but extending it to orchestrate more
complex compute operations in the medium term.  We need to make sure
choices made now are not in conflict with the medium term plan.

-- 
Russell Bryant



More information about the OpenStack-dev mailing list