Hi all, I’m hitting an error with dependencies and I was wondering if anyone knew a better way to dive deeper into debugging to figure out what dependency it is failing at? My solids are pretty simple:

@solid
def hello(_):
	return json.loads("{"msg":"hello"}")

@solid
def send_http_call(context, results):
	// Does an HTTP POST using the value from hello()

@pipeline
Def hello_pipeline():
	send_http_call(hello())

The actual error message is

Dependencies for hello_pipeline().compute failed

so trying to figure out how I can dive deeper to where .compute is occurring. Currently been reading up on the docs in case there’s a concept I missed

Hey @Edward Chen - do you mind including the full stack trace?

of course, thanks for the help!

2021-02-03 20:00:08 - dagster - DEBUG - hello_pipeline - e374b429-31fa-480a-9145-188594788e00 - 1446 - PIPELINE_START - Started execution of pipeline "hello_pipeline".
2021-02-03 20:00:08 - dagster - DEBUG - hello_pipeline - e374b429-31fa-480a-9145-188594788e00 - 1446 - ENGINE_EVENT - Executing steps in process (pid: 1446)
2021-02-03 20:00:08 - dagster - ERROR - system - e374b429-31fa-480a-9145-188594788e00 - send_http_call.compute - Dependencies for step send_http_call.compute failed: ['hello.compute']. Not executing.
2021-02-03 20:00:08 - dagster - DEBUG - hello_pipeline - e374b429-31fa-480a-9145-188594788e00 - 1446 - ENGINE_EVENT - Finished steps in process (pid: 1446) in 229ms
- 2021-02-03 20:00:08 - dagster - ERROR - hello_pipeline - e374b429-31fa-480a-9145-188594788e00 - 1446 - PIPELINE_FAILURE - Execution of pipeline "hello_pipeline" failed. Steps failed: ['hello.compute'].

very strange - it doesn't seem to be running the

hello

solid at all. how are you running the pipeline? with

execute_pipeline

? are you passing in a solid selection? what dagster version are you running?

I’m running it via docker containers so i ssh into the container and was able to figure out the reason. It looks like my original code had some issues with json decoding but was able to confirm that the issue is likely somewhere else. I assume the

__.compute

error is kind of a catch-all for exceptions if not specifically defined?

Glad that you figured it out. Yeah "hello.compute" is basically just the name of step that executes the hello solid. I think we may have removed that naming in a recent release, and now we just call it "hello"

Hi Edward - we're confused why the logs here didn't say something about starting the step before the failure (even if there was an error in the pipeline code) - there might be a problem with our error reporting somewhere. Is it possible to share more about what the problem ended up being / what your Dagster version is?

Absolutely, sorry that was my bad. Realized this was just the equivalent of “Solved.” on SO 😂 Version: 0.9.22.post0 Problem: I was getting a

PIPELINE_FAILURE - Execution of pipeline "hello_pipeline" failed. Steps failed: ['hello.compute'].

error message. I am running the dagster instance in a docker container and running it with

launchPipelineExecution

via GraphQL. Solution: There wasn’t more information from Postman so I had to ssh into the running docker container and run the curl

launchPipelineExecution

from there. That revealed that my post call was incorrect (I forgot the exact error but I believe it was something with json decoding)

No problem at all - so it was logging a useful error locally in the docker container, but nothing was appearing in the event log. Makes sense!