Re: [DISCUSS] Should we pre-install celery/k8s providers?

2023-07-24 Thread Ferruzzi, Dennis
Subject: RE: [EXTERNAL][DISCUSS] Should we pre-install celery/k8s providers? CAUTION: This email originated from outside of the organization. Do not click links or open attachments unless you can confirm the sender and know the content is safe. FYI. I will hold on with the formal vote until we have

Re: [DISCUSS] Should we pre-install celery/k8s providers?

2023-07-21 Thread Jarek Potiuk
FYI. I will hold on with the formal vote until we have a few more days for discussion. This is an easy change. J. On Fri, Jul 21, 2023 at 8:38 PM Daniel Standish wrote: > > Yeah it ain't the easiest of decisions. > > @niko, when I think about the staged approach, it feels like it is maybe >

Re: [DISCUSS] Should we pre-install celery/k8s providers?

2023-07-21 Thread Daniel Standish
Yeah it ain't the easiest of decisions. @niko, when I think about the staged approach, it feels like it is maybe more disruptive than doing it all at once. stage 1: force everyone to install k8s and celery libs -- that's one disruption / risk stage 2: remove pre-install -- another disruption it

Re: [DISCUSS] Should we pre-install celery/k8s providers?

2023-07-21 Thread Jarek Potiuk
Yeah. I thought a bit more about that and the risks of forcing Airflow 2.7 users to use the latest k8s provider even if they do not use K8S executor is much bigger (and much more difficult to handle) than risks of not installing the provider in the right version. Also I think there is one more

Re: [DISCUSS] Should we pre-install celery/k8s providers?

2023-07-21 Thread Mehta, Shubham
I second Jed's proposal and believe that the implications and benefits, as laid out, make a lot of sense. > we wouldn't have to set a minimum version for these 2 providers to ensure we > have the version with the executors, allowing an "old" k8s provider (KPO > itself is pretty common after

Re: [DISCUSS] Should we pre-install celery/k8s providers?

2023-07-21 Thread Oliveira, Niko
, 2023 10:07:44 AM To: dev@airflow.apache.org Subject: RE: [EXTERNAL][DISCUSS] Should we pre-install celery/k8s providers? CAUTION: This email originated from outside of the organization. Do not click links or open attachments unless you can confirm the sender and know the content is safe

Re: [DISCUSS] Should we pre-install celery/k8s providers?

2023-07-21 Thread Daniel Standish
* *provider* extras or provider optional features

Re: [DISCUSS] Should we pre-install celery/k8s providers?

2023-07-21 Thread Daniel Standish
Yeah I think this is an important point: > > Core Airflow so far hasn't had the dependencies > required to make those executors functional either - users either had to > use the extra or install the provider directly. So that doesn't really > change if we choose not to preinstall the providers. >

Re: [DISCUSS] Should we pre-install celery/k8s providers?

2023-07-21 Thread Collin McNulty
I like this solution conceptually. I have not considered every angle, but as long as there aren't any big gotchas, this seems like the right thing to do.

Re: [DISCUSS] Should we pre-install celery/k8s providers?

2023-07-21 Thread Jarek Potiuk
Yes. I think this makes much more sense. On Fri, Jul 21, 2023 at 6:19 PM Jed Cunningham wrote: > > Hello everyone, > > After thinking through whether the dask provider should be pre-installed or > not, it got me thinking about whether we should even be pre-installing the > celery and k8s

[DISCUSS] Should we pre-install celery/k8s providers?

2023-07-21 Thread Jed Cunningham
Hello everyone, After thinking through whether the dask provider should be pre-installed or not, it got me thinking about whether we should even be pre-installing the celery and k8s providers. Core Airflow so far hasn't had the dependencies required to make those executors functional either -