![]() ![]() Different worker and scheduler Pods appear as differently colored lines on this chart. If a given worker queue spins a worker down and back up again within a given interval, the newly spun up worker appears as a new color on the graph.ĬPU Usage Per Pod (%): This metric graphs the peak CPU usage for all workers and schedulers for a given interval. Hover over the graph to view a graph legend. Different worker and scheduler Pods will appear on these charts as differently colored lines. These metrics contain information about the Kubernetes Pods running your workers and schedulers. This metric can both help you understand how your pipelines are performing overall, as well as identify DAG/ task runs that didn't result in a failure but still took longer to run than expected. In the example above, the P90 Run Duration per Status for successful DAG/ task runs at 5:00 was 34 seconds, which means that 90% of those DAG/ task runs finished in 34 seconds or less. P90 Run Duration per Status: This metric graphs the 90th percentile of execution times for DAG/ task runs, plotted based on the DAG/ task run start time. To see timed out DAG runs, you must go into the Airflow UI to check on the statuses of each DAG run there. The DAG runs metric does not record DAG run timeouts as failed runs. Use this metric to see exactly when recent DAG/ task runs succeeded or failed. Runs per Status: This metric graphs the number of failed and successful DAG/ task runs, plotted based on the DAG/ task run start time. These metrics contain information about your Deployment's DAG runs and task runs over a given period of time.ĭAG/ Task Runs: This metric graphs the total number of DAG/ task runs. The following topics contain information about each available metric. To view metrics for a Deployment, open the Deployment in the Cloud UI and click Analytics. ![]() Because metrics are collected in real time, you can use this page to detect irregularities in your pipelines or infrastructure as they happen. This page includes metrics that give you insight into the performance of both your data pipelines and infrastructure. The Analytics page contains a suite of metrics for a given Deployment. This document explains each available metric and where to find them. ![]() They can also help you estimate the cost of your Deployments. These metrics are a useful reference as you troubleshoot issues and can inform how you allocate resources. The Cloud UI exposes a suite of observability metrics that show real-time data related to the performance and health of your Deployments. ![]()
0 Comments
Leave a Reply. |