Hi friends. Trying out deployment of dagster to k8...
# ask-community
m
Hi friends. Trying out deployment of dagster to k8s and having trouble getting past the Dagit readiness check probe at
/dagit_info
- when I check the pod-container logs there is no output at all. ๐Ÿงต thread here for some details
I've looked at the generated k8s manifest from my helm chart and see this:
Copy code
command: [
            "/bin/bash",
            "-c",
            "dagit -h 0.0.0.0 -p 8091"
          ]
I've tried to exec into a bash shell to see what's up in the container:
Copy code
kubectl exec -it dagster-dagit-zzzzzzzz -c dagster -- /bin/bash
when in the shell, manually starting dagit:
Copy code
root@dagster-dagit-zzzzzzzz:/# dagit -h 0.0.0.0 -p 8091
^ no output - it terminates with exit code 137
was that a memory OOM kill perhaps??
a
exit code 137
ya this generally indicates OOM
โœ… 1
m
๐Ÿ”ฅ good to know
let me try to give some more memory!
@alex do you know if there's a verbose/debug flag or something like that so that startup is less "quiet" during boot
looks like I'm able to get http 502s for readiness check at least now
(progress)
ahh nevermind, progress yet thus~
Copy code
Usage: dagit [OPTIONS]
Try 'dagit --help' for help.
Error: No arguments given and workspace.yaml not found.
I think I'm on my way ๐Ÿš€
a
๐Ÿ‘ no good debug/verbose flag at this time
technically you could manually run
uvicorn
and target the asgi app and then set whatever logging flags you want
๐Ÿ‘€ 1
uvicorn  <http://dagit.app:default_app|dagit.app:default_app> --factory
๐Ÿ‘ 1
m
is it possible to start dagit without any workspaces? was trying not to deploy the user-code example
I see the same problem with the daemon as well
a
dagit has an
--empty-workspace
flag, daemon i think doesnt support empty workspace
โœ… 1
๐Ÿ™ 1
d
Believe they use the same flags (but you'll want the daemon to use whatever workspace flags you end up doing for dagit once you have a workspace)
m
thanks @alex @daniel
condagster 1