https://dagster.io/ logo
s

szalai1

03/22/2021, 12:57 PM
Hi, we are upgrading to user deployments, question. As it's no longer possible to specify the executor image in the execution mode. Is there any way to specify different executor container on different pipeline from the same user code deployment ? (e.g data science deployment, executor with gpu driver, executor w/o drivers)
a

alex

03/22/2021, 2:36 PM
just to make sure we are on the same page - this is for celery k8s ?
s

szalai1

03/22/2021, 2:58 PM
yes
a

alex

03/22/2021, 3:02 PM
we had a prototype for some changes around this restriction cc @sashank
s

szalai1

03/22/2021, 3:49 PM
thanks, I know it was possible in 0.9.22 without user deployments I could just use in executor config the image I wanted
s

sashank

03/22/2021, 4:06 PM
hey @szalai1, this was supposed to go out a while ago but slipped. we’ll have the option go out in the next release to specify the job image in the executor config and have that take precedence over the image set in the user code deployment
👍 2
s

szalai1

03/22/2021, 4:08 PM
that's great, thank you very much 🙂
s

sashank

03/22/2021, 10:51 PM
@szalai1 we just shipped this in 0.11.1. let us know if you run into any issues
s

szalai1

03/23/2021, 7:18 AM
great, thanks a lot