I'm using self hosted retool. Currently on the free plan. I've deployed the service without temporal. I do have code-executor setup and configured in the backend. When I create a codeblock there is no option to switch to python.
I've setup temporal as well. But the service seems to think temporal is not setup.
I don't see any logs that can give me a hint on what exactly is happening.
I've all the 4 env variables configured
CODE_EXECUTOR_INGRESS_DOMAIN
WORKFLOW_BACKEND_HOST
WORKFLOW_TEMPORAL_CLUSTER_FRONTEND_HOST
WORKFLOW_TEMPORAL_CLUSTER_FRONTEND_PORT
I tried setting up locally and it works fine. But on ECS (where I upgraded the existing version it's not). Also, I noticed that there is no log that says
Creating new temporal client
in my ecs setup. any idea why it's not trying to connect to temporal even though the front end is configured?
I've solved the temporal issue as well. It had connection issues and also was expecting a namespace called workflows.
But now my code-executor fails with the following error: [E][2024-05-04T16:13:27+0000][24] standaloneMode():275 Couldn't launch the child process
Also, After sometime it crashes because of something related to datadog. /retool/code_executor/node_modules/dd-trace/packages/dd-trace/src/plugins/tracing.js:97 retool-code-executor May 04, 2024 at 21:40 (UTC+5:30) if (!span._spanContext._tags.error) { retool-code-executor May 04, 2024 at 21:40 (UTC+5:30) ^ retool-code-executor May 04, 2024 at 21:40 (UTC+5:30)TypeError: Cannot read properties of undefined (reading '_spanContext') retool-code-executor May 04, 2024 at 21:40 (UTC+5:30) at ChildProcessPlugin.addError (/retool/code_executor/node_modules/dd-trace/packages/dd-trace/src/plugins/tracing.js:97:15)
I’m seeing this same issue. The code-executor container starts, throws this same child process error a couple times, and stays running but retool doesn’t know it exists (ie. No python code blocks). Hopefully someone can help us shed some light on what this error indicates.