Automation triggering rules with huge delay
Platform Notice: Cloud - This article applies to Atlassian products on the cloud platform.
Summary
On some rare occasions, automation will take about an hour or more to execute an issue trigger resulting in executing the rule workflow with a huge delay, causing unwanted actions on certain issues matching the automation conditional blocks, making it look like the rule was triggered without any apparent reason.
Diagnosis
Some user instances have heavily loaded automations that run multiple times within short periods of time during the day. In normal circumstances, automation can keep up with a large number of automation executions in parallel without having a performance impact. However, there may be certain scenarios where multiple automations are running with thousands of parallel processes resulting in a degraded performance due to the high load. Automation uses a rule processing queue to manage the execution of rules in your instance. To maintain performance, rule executions are queued and the number of items processed in parallel is limited.
Cause
A large spike of automation executions can be seen under the Global automation performance insights page:
- https://<instanceName>.atlassian.net/jira/settings/automation#/performance-insights
Large automation spikes can be due to the following situations:
- Enabling automation that runs on multiple issues (Scheduled ones for example). Depending on the number of issues the automation takes action against.
- Restoring an archived project that has several issues and automation triggering against those. Also, having a high volume of automation rules running on other projects.
- Automation rules that are configured for looped "send web requests".
- Doing bulk imports can contribute to having a spike on the automation resources (specially for creating or updating issues).
The spike will be shown in the performance-insights graph, by filtering on the day the huge delay was seen, also listing the rules that contributed to the spike. For example:
Conclusion
This behavior is considered to be very uncommon and it should be attributed to isolated cases where huge volume of rules are being executed in parallel resulting on delays for certain issues where automation is running towards.
If you're facing recurrent delays due to automation spikes, or if you're experiencing delays not related with the documented here. Kindly get in contact with us here for better assistance.