Pipeline Execution per Instance
Learn more about what can trigger this alert.
Last updated
Learn more about what can trigger this alert.
Last updated
The Alerts feature is currently in restricted beta phase and is only available to specific customers. Learn more about the Beta Program.
Pipeline execution per instance is a metric that determines the number of executions per second of a pipeline per instance. If you use this metric to create an alert, you can set up a notification when the rate is outside the defined interval/range.
An instance (or replica) is the smallest part of a server where the pipelines are found. When you deploy a pipeline within the Run page, you can choose how many instances (replicas) you want to deploy.
When configuring an alert, you must specify these parameters:
Limit value: the execution rate to use as the limit for triggering the alert. This field accepts numeric values. Decimals must be separated with a period (.).
Condition: whether the alert should be triggered when the number of executions per instance per second is less than, equal or greater than the limit value for the specified time limit.
Tolerance limit: the period that the number of executions per instance per second must be outside the specified limit value before an alert is triggered. The alert is triggered only if the pipeline execution per instance remains within the specified threshold and period (between 5 and 20 minutes) as shown in the image below:
If the number of pipeline execution per instance is outside the expected range, it could be due to the following:
For example, if your pipeline has a REST trigger that is activated by a third-party application, a failure in that application will cause that pipeline not to activate. Consequently, the execution rates of your pipeline would take a value lower than the expected one. If this is the case, fixing the issue in the third-party application will return your pipeline execution rate to the expected level.
If, on the other hand, the execution rate of a pipeline is higher than expected, this may be due to an incorrect trigger configuration, such as a cron expression in the settings of a scheduler trigger that was entered incorrectly.
An improper pipeline architecture can lead to unexpected execution rates. For example, if you build a pipeline that reprocesses failed executions and forget to set a limit on reprocessing, that pipeline might try to execute an infinite number of times.
An inefficient architecture may also cause pipelines to take longer than expected to execute, decreasing their execution rate. If this is the case, review your pipeline architecture. Consider using an event-driven architecture and/or pagination.
Read the articles Event-driven architecture and Pagination to learn more about these concepts.
When the deployment size of a pipeline is too small to support the frequency in which it's triggered, execution requests stack up instead of executing immediately. This can cause the execution rate to take on a value that is lower than expected. If this is the case, consider increasing the deployment size of your pipeline. Read the article about deployment sizes to learn more.
For more information, read our Pipeline Metrics documentation.
