Starting work on Operator Working Group
Guillaume Demonet
+1 for me as well, thanks for the initiative!
On Mon, Mar 2, 2020 at 2:08 AM Peter Lunderbye <peter.lunderbye@...> wrote:
--
Guillaume DEMONET Scality - R&D Engineer +33 6 58 58 11 26 | guillaume.demonet@... http://scality.com/ | Twitter: @scality
|
|
Nicolas Trangez <nicolas.trangez@...>
+1 very interested as well. K8s' continuous refinement from high-level
state declarations to real-world realization, and Operators/CRDs to extend this model even further (including application/domain-specific knowledge translated to code) are what makes an investment in K8s compared to more traditional configuration management systems *really* worthwhile. Nicolas On Mon, 2020-03-02 at 11:52 +0100, Guillaume Demonet via Lists.Cncf.Io wrote: +1 for me as well, thanks for the initiative!
|
|
Naga Ravi Chaitanya Elluri
+1, I am in.
On Fri, Feb 21, 2020 at 11:04 AM Reitbauer, Alois <alois.reitbauer@...> wrote:
|
|
Felipe Musse
Great initiative, I'm very interested in participating.
|
|
Gerred Dillon
Alois, Are we ready to move to the next steps? The charter seems to be settling. I'd move to kick off an initial meeting, establish leadership, and establish the scope and times for this WG and begin reporting back to the SIG. Gerred
On Mon, Mar 2, 2020 at 12:01 PM Naga Ravi Chaitanya Elluri <nelluri@...> wrote:
|
|
james@...
I'd definitely be interested in being involved :) I'm especially keen to understand how we can:
* Define best practices around _deploying_ operators, including periphery components such as webhooks * Community guidelines on how to release and manage API versions: Kubernetes defines a lot of this already, but it is not surfaced clearly and is non-trivial * Guidelines on interoperability between operators, e.g. one third party extension depending on another (for context, in cert-manager we are often depended upon by other operators, which creates a 'dependency problem') Despite have '''opinions''' in the past, I'm actually less concerned about inventing a strict definition for "operator vs controller", and I'm not that convinced that defining the difference is actually beneficial to users. All of these things are 'controllers' in essence, and beyond that, I think 'operator' gets thrown around as a bit of a marketing term of the back of the original CoreOS blog post on this about the "etcd-operator". Thanks for putting this all together anyway, and looking forward to getting started :) (sorry for the dupe if anyone got one, the CNCF lists page has the "Group Reply" button to the right and it took me a good 5 minutes to spot it!!)
|
|
Scott Nichols
I am interested in participating in this SIG. I work on Knative and have spent a fair amount of time thinking about controllers and how to write them. I think the thing that no one has really talked about yet is the flavor and purpose of a controller/operator: I see three main types, one that extends functionality of K8s to do something new for you (like Knative Service) and a second type that is the operational instructions to keep a system running. A third is something that is reaching off the cluster to allow someone to write a declarative API for another component that might not be declarative. Lumping these all into the same name of "Operator" seems forced. I do like the Operator SDKs scales of an operator, and I think there might be a second axis around how broad that extension acts (its scope).
On Wed, Mar 4, 2020 at 7:53 AM <james@...> wrote: I'd definitely be interested in being involved :) I'm especially keen to understand how we can:
|
|
mhamdi.semah@...
I'd like to contribute. Looking forward to it.
|
|
Reitbauer, Alois
I am working on getting a meeting on the calendar next week.
From: <cncf-sig-app-delivery@...> on behalf of "mhamdi.semah via Lists.Cncf.Io" <mhamdi.semah=gmail.com@...>
I'd like to contribute. Looking forward to it.
|
|
Thanks Alois, let us know if there is anything we can help with.
Chris Hein
On Mar 13, 2020, 11:28 AM -0700, Reitbauer, Alois <alois.reitbauer@...>, wrote:
|
|