Structure - Looking for better ways to filter issues across levels (improve performance)

Dirk Evrard

Hi all,

I'm currently working in an on-premise JIRA with the Tempo Structure solution: overall it's really great, but I have build a traceability matrix to validate testing traceability in such a way that it leaves very to no room to change the hierarchy.
To give some context: it's a validated project with a very strong audit and quality management team, so while you may be tempted to suggest more efficient hierarchy levels, it's just not possible at this time to deviate from the structure design.

Let me start by showing you the relationships and issue types as they are defined in JIRA:

So basically I need to build a structure that needs to go 6 levels down, but the problem is that I NEED TO FILTER OUT any linkage where the TUS is not also linked directly with the CO through relationship "tests". In other words: I have HUGE performance issues because I need to load all the content (we're talking 1000's of lines), and only at level 5 can filter out through a formula like "CO_Link"

IF issuetype = "TUS": issueLinks.MAP(IF($.source = this, $.destination, $.source)) .FILTER($.issuetype = "CO")

This gives me the "root" CO, as through ancestry, I currently get ALL TUS which are somehow linked through other CO's taking every stair up.

Then I do another formula to check this outcome and return a "true" or "false" statement that I can use a filter on:

IF issuetype = "TUS" AND Parent#level=-4{issuetype} = "CO" AND CO_Link = Parent#level=-4{key}: "TRUE"

At this point the structure used to work perfectly, but these days it's breaking down because we are talking thousands of issues as the scope continues to increase.

I dabbled in S-JQL by using a "child of root" filter (as the TUS is linked directly to the CO), but this failed to work, as it only considers hierarchy levels in the structure, and doesn't seem to work when it's 5 levels deep. If I use "descendant of root", I keep getting the TUS I don't need, as they are indeed all descendants, but just not directly linked through the green arrow in my screenshot...

Overall I've tried a million things at this point (because you also want to filter out levels 2 to 4 if there is no level 5 match.

So I'm wondering:
- Is this problem even solvable in JIRA structure, maintaining the mandated linkages.
- Is there any way I can improve performance so the structure loads fast (it currently exceeds 600 seconds and times out).

A big thank you for your advice.

Stepan Kholodov

Hello @Dirk Evrard 

An alternative approach can be considered here:

- start building the structure by adding the right TUS issues. If you can build a JQL query that returns the issues you need in Jira, then you can add the same scope of issues in the structure with an Insert generator.
- add the test Executions underneath TUS issues using an extender.
- add the levels above by using the Issue Links Group generators, one generator per each level (the 'Consider other groups' option should be enabled in the configuration of the groupers). You might need to rearrange the groupers to achieve the right hierarchy order.

As for the performance of your current structure - it is possible that something can be done to improve it. For example, limiting the Extend on level setting of all Extend generators to only appropriate levels of the hierarchy can speed things up quite a while. But it depends on the details of your setup.

In any case, we'll be glad to look into it, and see what can be done. Please reach out to us directly at our support portal, and we'll go from there.

Best regards,
Stepan
Tempo (the Structure app vendor)