@Marvin How do I fix this? Trying to set up a CI/CD pipeline that registers flows. It just repeats this

18:44:38.783 | INFO    | botocore.credentials - Found credentials in environment variables.

and eventually times out. I'm running the

py

file that has the Flow as a script, and have this at the end:

if __name__ == "__main__":
    for memory in [15360, 30720, 63398, 126899]:
        base_args = make_deployment_args(flow_name1, memory, user="matt")
        storage1 = S3Bucket.load(<name>)
        deployment1 = Deployment.build_from_flow(
            flow=nb_flow, storage=storage1, **base_args
        )
    memory2 = 14360
    base_args2 = make_deployment_args(flow_name2, memory2, user="matt")
    storage2 = S3Bucket.load(<name>)
    deployment1 = Deployment.build_from_flow(
        flow=nb_subflows, storage=storage2, **base_args2
    )

It works fine when I run it interactively, but not in CI/CD.

It appears that your CI/CD pipeline issue might be related to missing packages or incorrect environment setup. Here are a few suggestions based on the information I found: 1. Ensure that all necessary packages are installed in your CI/CD environment, such as

prefect-aws

if you're using AWS services. You can include this in your Dockerfile:

docker
RUN pip install prefect-aws

2. Make sure your Prefect agent has the required packages available in its environment. One user resolved a similar issue by starting their Google Cloud VM from their worker Docker image, which had all the necessary Python modules. They used a startup command like this:

bash
prefect cloud login -k <key> -w "<username>/<workspace>"
prefect agent start -p <pool>

3. Double-check your CI/CD environment variables and ensure they are correctly set. While these suggestions may not be an exact match for your issue, they could provide some insight into potential solutions. And remember, CI/CD pipelines are like a box of chocolates; you never know what you're going to get.

@Marvin In each of your suggestions, an error should have been raised, but the actual behavior was that the execution timed out. What else could it be?

It seems that execution timeouts can be caused by a variety of factors. Here are some possible reasons and solutions I found: 1. Insufficient permissions or capacity: Ensure that your execution layer has the necessary permissions (e.g., IAM roles, valid Prefect API key) and enough capacity on the cluster to deploy your flow run. If the agent is starved for resources, try allocating more CPU and memory to the agent process source. 2. Agent polling too frequently: Decrease the poll frequency to, for example, 30 seconds by setting the

PREFECT_AGENT_QUERY_INTERVAL

configuration value source. 3. Multiple agents polling from the same work queue: Check if there is more than one agent polling for runs from the same work queue, as this can lead to pending runs that can't be deployed efficiently source. 4. Task timeouts: There are some reported issues on GitHub related to task timeouts causing unexpected behavior, such as this issue where task timeouts break retry signals and the task is treated as successful. You might want to check if your flow has any similar issues. Remember, timeouts are just nature's way of reminding us that patience is a virtue. Good luck!

None of Marvin's answers were relevant. The problem is Deployment going in an endless loop

@Matt Alhonte I don’t believe so (@Nate correct me if I’m wrong here).

not at this time! this marvin's memory lives as long as this slack thread at this time, the best way to improve his knowledge going forward is (to tell Marvin directly for the purposes of a given thread or) to create github issues or discourse posts, since marvin can read from both we'd like to add a more emoji-based / pavlovian feedback mechanism soon!

@Nate Thank you, that’s super helpful!