Short answer: no. A task doesn’t need to return anything to show up in the UI. What matters is: - It’s decorated with

@task

and executed within a

@flow

- The flow run is connected to a Prefect API (Cloud or a running Server), so the run is tracked If a task returns nothing, it will still appear with state, logs, and timing. You just won’t see a “result” value unless you return one and (optionally) persist results. Tip: - Use

get_run_logger()

inside tasks to surface useful info in the UI logs. - If you want something richer than logs, create artifacts (e.g., markdown) during the task. Example:

from prefect import flow, task, get_run_logger
from prefect.artifacts import create_markdown_artifact

@task
def do_side_effects():
    logger = get_run_logger()
    <http://logger.info|logger.info>("I did a thing.")
    create_markdown_artifact("# Task report\nAll done ✅", description="Run notes")
    # no return

@flow
def my_flow():
    do_side_effects()

if __name__ == "__main__":
    my_flow()

If you want task return values visible across runs, configure result persistence (e.g., set result storage and/or

PREFECT_RESULTS_PERSIST_BY_DEFAULT

): https://docs-3.prefect.io/v3/advanced/results If you’re not seeing your task in the UI, are you running it inside a flow and against a Prefect API?

Good question! Do you mean: - Monitoring the Postgres database that backs your Prefect Server with Prometheus, or - Exposing Prefect runtime metrics themselves to Prometheus? Also, are you on Prefect Cloud or self-hosting Prefect Server? Short answers: - Self-hosted Prefect Server: Yes — you can monitor the Prefect Postgres database with Prometheus using the Postgres exporter. - Prefect Cloud: You can’t connect to the managed database directly. For metrics, you’d either instrument your own code or rely on Prefect’s built-in observability and alerts. - Prefect runtime metrics: Prefect doesn’t expose a first-class Prometheus /metrics endpoint. If you want Prometheus metrics about your flows, you can instrument your code with the Prometheus client or export infra metrics (e.g., Kubernetes, Docker) and correlate with Prefect events. If your goal is database monitoring (Postgres) with Prometheus - Use the Postgres exporter and point it at your Prefect DB. - Create a read-only DB user for safety. Links: - postgres_exporter - Prometheus scrape_config docs - Bitnami PostgreSQL Helm chart: metrics - Grafana PostgreSQL dashboard (ID 9628) Example docker-compose snippet

services:
  postgres:
    image: postgres:16
    environment:
      POSTGRES_DB: prefect
      POSTGRES_USER: prefect
      POSTGRES_PASSWORD: yourpassword
    ports: ["5432:5432"]

  postgres-exporter:
    image: <http://quay.io/prometheuscommunity/postgres-exporter:v0.15.0|quay.io/prometheuscommunity/postgres-exporter:v0.15.0>
    environment:
      DATA_SOURCE_NAME: "<postgresql://prefect:yourpassword@postgres:5432/prefect?sslmode=disable>"
    ports: ["9187:9187"]
    depends_on: [postgres]

Prometheus scrape config

scrape_configs:
  - job_name: 'postgres'
    static_configs:
      - targets: ['postgres-exporter:9187']

Kubernetes (Helm) option - If you use Bitnami’s PostgreSQL chart, enable metrics:

--set metrics.enabled=true \
--set metrics.serviceMonitor.enabled=true

- Or deploy prometheus-postgres-exporter Helm chart and provide a connection secret. If your goal is “Prefect metrics in Prometheus” - There’s no built-in /metrics from Prefect 3 components today. - Two common patterns: - Instrument your flow code with the Prometheus Python client and either expose an HTTP endpoint from your runtime or push to a Pushgateway. - prometheus_client and Pushgateway - Rely on infrastructure metrics (K8s/Docker/EC2) in Prometheus and use Prefect’s event/alerting system for run-level status. If you can share: - Cloud vs self-hosted - Where your Postgres lives (container, managed service, K8s) - What specific metrics you want (DB health vs flow-run success rates) …I can give a targeted config for your setup.