Slow loading of large Jira filter list(~48k filters) when using Great Gadgets in Confluence

Issue #92 resolved
C-George.Sagen created an issue

When trying to use many Great Gadget macros in Confluence, the filter list takes around 50 minutes to load. The Jira instance it is pulling from with an application link has around 48k filters.

When I go to edit or create a gadget that has a filter list, it starts loading, and will update the filter count once it gets its first chunk. This means it can be showing just a loading spinner for 2-3 minutes before it gets its first chunk, which is long enough to make some users wonder if it is stuck. Once it gets the first chunk, then it shows progress like “Filters loaded: 1600”, which is good. Here is an example of a few times I’ve checked:

Filters loaded / seconds taken

11200 / 800s
14400 / 1200s
17600 / 1350s
20800 / 1550s
30400 / 1700s
41600 / 2600s

After loading, sometimes when clicking the dropdown of filters, it will open up a blank dropdown. Generally it doesnt do this, but its pretty rough when youve waited an hour and then the list still doesnt show in the browser.

It doesnt seem to cache the filters, as I can load the filters successfully once after 50 minutes, then edit again after a minute, and it will have to reload all over again.

Some gadgets that experience this issue:

  • Work Break-down Structure (WBS) Gadget
  • Cycle Time Trend Gadget
  • Control Chart Gadget
  • Filter Based Formula Gadget

Jira:
Data Center 8.13.2

Confluence:
Data Center 7.4.1

Great Gadgets for Jira Server:
1.21.2

Please let me know if there is a way for these filters to load at a reasonable speed, or with caching, or without getting a blank dropdown.

Comments (7)

  1. Danut M [StonikByte] repo owner

    Hi George,

    Thanks for posting. We are currently investigating.

    Do you have similar loading time in Jira? Or it is just in Confluence?

    Danut

  2. C-George.Sagen Account Deactivated reporter

    Hello Danut,

    When I try the same gadget, for example Cycle Time Trend, in the Jira instance that our Confluence instance is connected to through an application link, the load times for those filters are between 2 and 6 minutes. We’ve tried in some of our lower dev and test setups of Jira and Confluence, to see if the network in our production instance is just slow, but we had the same results where Jira loads relatively quickly for the number of filters, but Confluence takes 45 minutes or more.

  3. Danut M [StonikByte] repo owner

    Thanks George.

    We are evaluating the possibility of replacing the filters dropdown with a filter picker. This should allow searching for filters while typing their name. This way it will not get the whole list of filters from Jira but a limited number of matching filters. This should solve the issue.

    I will keep you updated with our progress.

    Danut

  4. Danut M [StonikByte] repo owner

    Hi George,

    We made a fix by replacing the filters dropdown with a filter picker.

    Please install version 1.21.3 available here:

    https://bitbucket.org/StonikByte/great-gadgets-add-on/downloads/great-gadgets-server-plugin-1.21.3.jar

    To install it:

    • Download the jar file from the provided link
    • Log on to Jira as an admin. Use the main Jira URL of your datacenter cluster (not a specific Jira node URL)
    • Go to Jira Administration > Apps > Manage Apps
    • Click Upload app and install the downloaded jar file

    Please let me know if this solves the problem. Once we’ll have your confirmation we will officially release the version on Marketplace.

    Thank you,

    Danut Manda

  5. C-George.Sagen Account Deactivated reporter

    Hello Danut,

    Thank you for the incredibly fast turn around on this! Preliminary testing has shown this to be an excellent fix - Super snappy, taking less than 30s even against our large list of filters. Our users will know the name of the filter they plan to use anyway, so no need to scroll through a list when they can just search it up!

    We will do a bit more testing and see if we can get this deployed to our production instance tonight, otherwise I will provide feedback for any issues we come across.

  6. Log in to comment