Hello,

Fernanda Reple (freple@atlassian.com) requested us to create a ticket here for tracking purposes regarding our ongoing discussions around developing custom Xray JQL Functions for Jira Cloud. We have already been in contact with Atlassian engineers on the matter. A summary below:

The original request can be found here, however, after initial discussions we realize this request is not possible to achieve: https://ecosystem.atlassian.net/servicedesk/customer/portal/34/ECOHELP-62263?created=true

Instead, our new request is to reduce the cache TTL from its current 7 days duration to something smaller like 1 day duration.

Reason: this will drastically reduce the amount of precomputations that need to be updated during the day, and will ultimately result in a much smaller time window where our customer's JQL functions are out of sync with the live data in their Jira instance.

Customer use case & explanation: Currently, if users execute JQL functions, the results get stored in the cache for the following 7 days and any time the function is executed afterwards the results come from the same cache. During this time it is up to Xray to update the results of the cache. Due to performance issues, we cannot update every unique JQL function + parameter set any time that an underlying Jira Issue is updated. Instead, we created a nonstop background job to update all precomputations. As soon as all are updated, it starts again from the beginning. The problem is this process can take anywhere from 20 minutes to a few hours to complete and restart, meaning users will potentially be seeing results from JQL queries that are up to a few hours outdated. This is extremely problematic for a reporting feature and for customers who need live information about time-sensitive release activities, especially for hotfixes. The request to decrease the TTL to 1 day will drastically reduce the number of precomputations that need to be updated in each cycle of the background job, thus decreasing its time to completion and time window where customers might see outdated data.