Hi folks - my team is looking into adopting prefect 2.1. One of the functionalities that we need is to be able to run a flow every time some external parameters change (if no flow is being run, run the flow, if a flow is being run, wait until it finishes). I was looking into the concept of work_queues and it satisfies that need mostly - however, we only care about running the latest job in the queue (the one that has the most up to date parameters) is there a way for the agent to always pick the most updated job in the queue and discard the others?

Hello! is there a way to add tags to a flow run created with

prefect.deployments.run_deployment

?

Hello - I have a weird one. I’m hosting an orion server locally via docker, I have an external volume attached but the volume directory is empty. I was expecting a sqlitedb and config files?

Hello, I am migrating from v1 to v2, in my old codebase, I use

ShellTask

to run some command

ShellTask(stream_output=True, slug="xxxx", timeout=3600).run(command=xxxx)

Is it deprecated in v2 world? Everything is covered by

@task

Hi, I deploy a service with the following CONFIG in GKE, but I get the following error in INGRESS.

some backend services are in unknown state

What is wrong? CONFIGFILE

apiVersion: batch/v1
kind: Job
metadata:
  name: flow-run-job
  labels:
    app: flow-run-job
spec:
  template:
    spec:
      containers:
      - env: []
        name: prefect-job
        image: image-path
        command: ["/bin/sh", "-c"]
        args:
          #deployments
          - python ./workflows/tutorial.py;
            prefect agent start -q test;
            prefect orion start --host 0.0.0.0 --log-level WARNING;
        ports:
        - containerPort: 4200
      restartPolicy: Never
---
apiVersion: v1
kind: Service
metadata:
  name: flow-run-job
  namespace: default
  labels:
    app: flow-run-job
spec:
  ports:
    - port: 4200
      protocol: TCP
  selector:
    app: flow-run-job

Hi prefect community. I am trying to handle a known timeout issue, but I can’t seem to be catch it using try; except. Here is the toy case that shows the problem:

from prefect import flow, task
from time import sleep

@task
def sleep_task():
    sleep(5)
    return 'sleep_task complete'
    
@flow(timeout_seconds=1)
def timeout_flow():
    print(sleep_task())
    return 'timeout_flow complete'
    
@flow
def main_flow():
    try: 
        print(timeout_flow())
    except Exception as e:
        print('error caught:: ', repr(e))

The try; except block does not catch the error, instead the framework throws a TimeoutError exception. How can I catch prefect timeouts?

Hey there! Could it be that the Prefect Cloud or at least the Google SSO is currently down? Whenever I try to login, I am greeted with a blank screen and the following error pops up in the Console:

Failed to load module script: Expected a JavaScript module script but the server responded with a MIME type of "text/html". Strict MIME type checking is enforced for module scripts per HTML spec.

. I am running Chrome on Mac. A colleague of mine who is on a different network has the same issue by the way. Thanks 🙂

Hi team - having trouble signing in with the new magic codes..

hi everyone! i have one silly doubt. when we run a orion server locally how do we access the ui . http://127.0.0.1:4200 this just give me 404 also can we run prefect orion on k8s and use that to create infra blocks, flow deployments as well as run deployments ?

Hello, If I have a flow with DaskTaskRunner, but some tasks have parameter dependencies between them, and some other tasks have specified dependency using "wait_for", so essentially the tasks in the flow can be divided into a group of 3, and each group can run in parallel. DaskTaskRunner recognizes that right? I just want to make sure if having any dependency in there affects parallelism even for tasks that could run in parallel? I didn't want to separate them into 3 subflows because the logic is better as one flow, but wondering if I have to.

Hello, we have had a flow scheduled every night that has run successfully, but two nights ago it did not submit to the prefect agent and we are not sure why (no configs have been changed). Is it possible to use the audit trail to see if/when a scheduled flow is submitted to the agent, or any information around the submission of scheduled flows?

Hi there, I could not find the guide to install Prefect Orion Server on my on-premisses Kubernetes. Does anyone has a good tutorial on how to do it?

Hi. I am getting weird error when trying to load String block. It works fine locally, but in Github action fails with the error in thread.

Anyone else having outages with the Prefect 2.0 UI?

Hi! Do you have a recommendation how to use prometheus to observe task and flows, or is it a bad idea ?

Hi All. Is there a way to see the final context of a flow? Can I call prefect.context.something and pass in the flow id to get info after its completed?

FYI, Slack services are having an issue with threads.

I’m at a dead end. I really need to get Prefect running in Django. If anyone has any resources please share them… So far everything I have tried has failed. At this point I’ll re-try anything.

in prefect 1, how to return a value from a flow? im calling flows via

create_flow_run

and i want to pass what i generate in the flow to a subsequent flow

What is the prefect API URL to log in as a service account? (It only seems to give the key)

hi team! is there a plan to add something similar to the

Projects

abstraction from Prefect 1? I found it useful to be able to group flows and check system dashboard per project. Or would you recommend to just use tags and custom views instead? (which is only possible with flow runs, not with flows or deployments)

Hi, I am starting with prefect and stuck in a point. I am migrating my scraping system in prefect where I have to run my tasks in different remote servers. As per the sample flow attached, for the sake of distribution I need my second task to run in two remote servers in parallel. Can someone help me with right resource to refer?

Does anybody have any working examples of RRuleSchedules starting at a half past? So for example I want a flow to run every 5 minutes, between 10:30am and 4pm, Monday to Friday. My first thought is to use a rruleset but so far I haven't been able to successfully stringify

I have a main flow with multiple sub flows, where each sub flow is expected to run one after the other. Ideally if one of them fails, the rest shouldnt execute and that was how it was. However, when I pass the argument "return_state=True" to the subflows, it continues the execution of the next flow even if the previous one has failed. Is there a workaround with this?

Hello experts~ marvin I'm using prefect 1.x and Cloud. I made it to run a certain flow registered in the cloud with the GraphQL API. In this time i want to use the flow run_name in that flow. Every flow automatically creates a run_name as it is executed, right? Can't we put in the a task inside that flow which gets the flow's run name that is executed by API? There is only one flow here, and I asked because I wanted the output of that flow to be specified as the flow's run_name.

hey prefect. We are using v1 right now and the PG database is getting pretty large fast and appears to impact performance. Do you see any issues with archiving the database tables? Such as the flow, log, message, flow_run, flow_run_state tables?

Hello folks, I'm trying to migrate from v1 to v2, and I'm currently trying to figure out what to do with

apply_map

. Is is correct to assume that I can easily change my apply_map uses, by subflow runned in parallel?

Hi all, I am currently trying to make a dbt block in prefect 2 but whenever I register it through the CLI I get

Successfully registered 0 blocks

and the block is not created in the Cloud UI. I have been following the instructions given in the collections documentation and the prefect-dbt github

Hi all - we’re noticing some strange behavior on flow run links that are being generated in Prefect Cloud. Some flow runs have a URL pattern of:

<https://app.prefect.cloud/account/ACCOUNTID/workspace/WORKSPACEID/flow-runs/flow-run/FLOWRUNID>

while others use:

<https://app.prefect.cloud/account/ACCOUNTID/workspace/WORKSPACEID/flow-run/FLOWRUNID>

Where the first example has

flow-runs/flow-run

and the second just uses a

flow-run

format. Some of our users get a 404 error on all links of the first type, and others get 404 on all links of the second type. Does anyone know what might be happening or how to fix this?

Hi, I was wondering if there are any plans for supporting nomad by hashicorp as a kubernetes alternative.