https://prefect.io logo
Join Slack
Powered by
Hi. I’m trying Prefect Orion. I tried to run deplo...
# ask-community
m
Hi. I’m trying Prefect Orion. I tried to run deployment. However, the flow status stays ‘pending’ and is not executed. Does anyone have any idea what the reason might be? I would be happy to know.
The following steps were executed.
Copy code
--------

# Prefect orion start.
% prefect orion start

 ___ ___ ___ ___ ___ ___ _____    ___  ___ ___ ___  _  _
| _ \ _ \ __| __| __/ __|_   _|  / _ \| _ \_ _/ _ \| \| |
|  _/   / _|| _|| _| (__  | |   | (_) |   /| | (_) | .` |
|_| |_|_\___|_| |___\___| |_|    \___/|_|_\___\___/|_|\_|

Configure Prefect to communicate with the server with:

    prefect config set PREFECT_API_URL=<http://127.0.0.1:4200/api>

View the API reference documentation at <http://127.0.0.1:4200/docs>

Check out the dashboard at <http://127.0.0.1:4200>
INFO:     Started server process [26251]
INFO:     Waiting for application startup.
INFO:     Application startup complete.
INFO:     Uvicorn running on <http://127.0.0.1:4200> (Press CTRL+C to quit)
...

--------

# Build deployment.
% poetry run prefect deployment build ./workflows/flow_hello.py:main_flow -n flow_hello -t test
Found flow 'flow_hello'
Deployment YAML created at '/********/********/********/********/main_flow-deployment.yaml'.

--------

# Set 'PREFECT_API_URL'.
% prefect config set PREFECT_API_URL=<http://127.0.0.1:4200/api>
Set 'PREFECT_API_URL' to '<http://127.0.0.1:4200/api>'.
Updated profile 'default'.

--------

# Apply deployment.
% prefect deployment apply main_flow-deployment.yaml
Successfully loaded 'flow_hello'
Deployment 'flow_hello/flow_hello' successfully created with id '04a098ba-b53b-4216-98a2-9c96b059104f'.

--------

# Start Prefect Orion Agent.
% prefect agent start -t test
Created work queue 'Agent queue test'
Starting agent connected to <http://127.0.0.1:4200/api>...

  ___ ___ ___ ___ ___ ___ _____     _   ___ ___ _  _ _____
 | _ \ _ \ __| __| __/ __|_   _|   /_\ / __| __| \| |_   _|
 |  _/   / _|| _|| _| (__  | |    / _ \ (_ | _|| .` | | |
 |_| |_|_\___|_| |___\___| |_|   /_/ \_\___|___|_|\_| |_|


Agent started! Looking for work from queue 'Agent queue test'...

--------

# Run deployment.
% prefect deployment run flow_hello/flow_hello
Created flow run 'cerise-adder' (fc7e440d-05cb-4228-83a0-de7f70015531)

--------

# After execute command 'prefect deployment run flow_hello/flow_hello', output folling logs by Prefect Orion Agent.
22:27:59.810 | INFO    | prefect.agent - Submitting flow run 'fc7e440d-05cb-4228-83a0-de7f70015531'
22:27:59.877 | INFO    | prefect.infrastructure.process - Opening process 'cerise-adder'...
22:27:59.884 | INFO    | prefect.agent - Completed submission of flow run 'fc7e440d-05cb-4228-83a0-de7f70015531'

--------
t
Hi Masato, you have each one running in a separate terminal right?
m
Yes.
t
Strange. Looks like you build your deployment with tag 
test
and started an agent using the same tag so they should work.
Somehow the agent isn’t picking up on the flowrun
Wonder if it might be poetry somehow?
m
Strange. Looks like you build your deployment with tag test and started an agent using the same tag so they should work.
I think so too. I created a new queue and tried it. However, it didn’t work…
Copy code
prefect agent start 7a9b41ee-1e94-4f9d-b86f-865289ed0433
t
I’m trying to find any discourse discussing poetry and coming up short 😕 (though tbh i could have looked harder so don’t let that stop you from further digging 😅) I wonder if its worth building the same test with a venv or conda environment? I’ve just heard people struggling with poetry in the past.
I’m sure poetry is doable, but just might be useful to isolate that as the problem.
m
I tried with venv. However I got the same result...
t
Have you tried running the flow from the UI?
Going to Deployments -> test_flow/test_deployment -> clicking the run button on top right
m
Yes. Tried both UI and CLI.
t
Maybe creating a work-queue and using the work-queue id in the 
agent start command
instead of a tag?… I am stumped, but ultimately looks like things are working just your agent and work-queue aren’t talking properly for some reason. Sorry I can’t be of more help. I’ll flag this to Anna, see if she notices anything I couldn’t when she gets back.
Double check that the terminal where you are running the agent, when you run 
prefect deployment ls
you see your test deployment listed. Same with work-queue.
Also might be an issue coming from profile somehow. Sometimes I reset that file by just deleting it 
~/.prefect/profiles.toml
.
m
Maybe creating a work-queue and using the work-queue id in the 
agent start command
instead of a tag?
…
Sometimes I reset that file by just deleting it 
~/.prefect/profiles.toml
.
I got the same result...
Double check that the terminal where you are running the agent, when you run 
prefect deployment ls
you see your test deployment listed. Same with work-queue.
I tried it. There seems to be no problem.
I am stumped, but ultimately looks like things are working just your agent and work-queue aren’t talking properly for some reason. Sorry I can’t be of more help.
Don’t worry. I thank you for your help.
a
We've seen several people having import issue using venv, can you try with Conda instead?
Alternatively destroy your current venv and create a new one from scratch, activate it before starting the agent, same for the terminal from which you create deployment
m
Thanks. I will try it. I noticed that it seems to be slower than it should be until the flow is executed. After running the deployment run, it runs 30 minutes ~ 1 hour later, and the status changes to ‘Completed’.
a
yup, changing to conda should fix it for you
m
Hi. Unfortunately, I was not able to solve this problem using conda. I finally solved this problem by using this Container Image It’s a little unclear, but I’ll try to work this way. https://hub.docker.com/r/prefecthq/prefect Thank you very much for all your advice.
panda dancing 1
t
So glad you got it working and thank you so much for following up Masato!
🙏 1
37 Views
Open in Slack
PreviousNext
Powered by