[GitHub] [airflow] potiuk commented on issue #24914: Metric grouping and task filtering

[GitBox <gi...@apache.org>]

potiuk commented on issue #24914:
URL: https://github.com/apache/airflow/issues/24914#issuecomment-1181414773

   > Thanks for getting back to me so quickly! Normally when I'm looking at these charts I'm looking for things like anomalous run times (either becuase of starting late or taking too long) and degradation of performance... general indicators that we need to take some corrective action on the pipeline to keep it hitting SLA. Ideally we would also use the SLA functionality in Airflow but in its current format it's not particularly useful. Until then we're reliant on data presented in the UI itself with someone monitoring it on a periodic basis.
   
   I think it's actually best if you can monitor your Airlfow SLA for Tasks outside of Airflow. The SLA feature is indeed not very useful, but I personally think Airlfow in it's future incarnation will become more of the platform and expose more of it's metrics and stats to external systems that are way better in "monitoring" than Airflow will ever be. Usually I'd recommend to integrate stats from Airflow into other monitoring tools you already use.  This will be even easier after we add open telemetry support: https://cwiki.apache.org/confluence/display/AIRFLOW/AIP-49+OpenTelemetry+Support+for+Apache+Airflow
   
   A very good example of what can be achieved even today was actually presented at the Airflow Summit 2022 https://airflowsummit.org/sessions/2022/the-slayer-your-data-pipeline-needs/. - this is a cool example of how you can build your monitoring using Airflow as Platform.


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: commits-unsubscribe@airflow.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org