Hi All, I have deployed dagit on open shift. but i...
# ask-community
g
Hi All, I have deployed dagit on open shift. but its erroring out with "CrashLoopBackOff" as Pod status. In the logs I see "*Welcome to dagster! you must pass the application as import string to enable 'reload' or ;workers' "*
p
This looks like a uvicorn-generated message, and potentially unrelated to the crash loop? What version of dagster/uvicorn are you running?
g
@prha I dont have uvicorn in my project, dagster is 0.15.1
I have 0.18.1 uvicorn but i dont think its ever used.
p
dagit uses uvicorn under the hood… is it possible to create a pod and try to start dagit manually?
g
I have main.py in which from dagit import cli as dagit_cli dagit_cli.main() To start the dagit on open shift
@prha not sure how to do it manually
p
Are you able to spin up a pod and the exec the
dagit
command on it?
g
so the issue with my set up is that, we dont want to pip install dagster dagit. Then run "dagit".
we want to run the dagit using a py file.
from dagit import cli as dagit_cli
dagit_cli.main()
p
Yeah, that should still work. I’m still confused as to why
uvicorn
is complaining about the application…
Here’s our only call to uvicorn, which should only error if we pass in
reload
or
workers
config params, which we are not: https://github.com/dagster-io/dagster/blob/b71f15031ad3e869065b4d0ce45d5fb8d54104ab/python_modules/dagit/dagit/cli.py#L168
g
from dagit import cli as dagit_cli
import sys, os
sys.argv = [sys.argv[0], "--empty-workspace"]
dagit_cli.main()
does that cause any issue ?
p
That should not cause any issues, no
g
😢, not sure what is causing this issue to happen on the pod
p
Yeah, the error message doesn’t jibe with my understanding of what the code for your dagit/uvicorn versions are doing. Do you have high confidence that you are running
dagit==0.15.1
and
uvicorn==0.18.1
in those pods? Do you run into any issues when you try to run your setup locally (i.e. not in open shift?)
g
dagit == 0.15.5 and uvicorn[standard] == 0.18.2
no I dont see any issue when ran locally