Uploaded image for project: 'Jira Data Center'
  1. Jira Data Center
  2. JRASERVER-66067

Gadget for Two Dimensional Filter Statistics may cause high memory pressure in case of large dimension

XMLWordPrintable

    • 7.02
    • 32
    • Severity 2 - Major
    • 39
    • Hide
      Atlassian Update – 09 September 2024

      Thank you for reporting this issue. In the last weeks we have been working hard on fixing it.

      Summary of the problem:

      The gadget for two dimensional filter statistics in case of large dimensions was using significant amounts of heap, leading to memory pressure (potentially GC pauses, OOM and the like). The problem was caused by creating large in-memory list of results prior to sending a response to the client. The second cause was allocating excessive amounts of comparator objects.

      New behaviour after the change:

      After the fix, the data will be sent to the client immediately therefore reducing the memory usage server-side. The excessive allocations have also been fixed. The server-side code should not be using excessive amounts of heap anymore. 

      Status of the fix and Fix Version:

      The fix is ready, and we’re moving the status of this ticket to Waiting for release with Fix Version 10.1.0 and 10.0.2.

      We're actively working on having this fix released on 9.12.14, scheduled for 2 October 2024

      Best regards,

      Bartosz Radaczyński
      Jira DC Principal Software Engineer

      Show
      Atlassian Update – 09 September 2024 Thank you for reporting this issue. In the last weeks we have been working hard on fixing it. Summary of the problem: The gadget for two dimensional filter statistics in case of large dimensions was using significant amounts of heap, leading to memory pressure (potentially GC pauses, OOM and the like). The problem was caused by creating large in-memory list of results prior to sending a response to the client. The second cause was allocating excessive amounts of comparator objects. New behaviour after the change: After the fix, the data will be sent to the client immediately therefore reducing the memory usage server-side. The excessive allocations have also been fixed. The server-side code should not be using excessive amounts of heap anymore.  Status of the fix and Fix Version: The fix is ready, and we’re moving the status of this ticket to Waiting for release with Fix Version 10.1.0 and 10.0.2 . We're actively working on having this fix released on 9.12.14 , scheduled for 2 October 2024 .  Best regards, Bartosz Radaczyński Jira DC Principal Software Engineer

      Summary

      JIRA Gadget for Two Dimensional Filter Statistics may cause high memory pressure in case it was built on a large data set and large dimensions were used.

      Environment

      • Large JIRA instance

      Steps to Reproduce

      1. Create JQL which returns a large data set
      2. Add Two Dimensional Filter Statistics to Dashboard and choose same parameter for X/Y axis. Parameter should have many values: eg Assignee, Versions, etc
      3. Save gadget
        • If you press Show more you can see more results, which leads to high memory consumption.

      Expected Results

      JIRA performs well. If required it will limit the results set.

      Actual Results

      JIRA uses large amounts of memory 4GB+ which can lead to OOM.

      Notes

      None

      Workaround

      None

        1. TwoDimensionalFilter_example.png
          64 kB
          Andriy Yakovlev [Atlassian]

              aea6bfd9e5b3 Bartosz Radaczyński
              ayakovlev@atlassian.com Andriy Yakovlev [Atlassian]
              Votes:
              12 Vote for this issue
              Watchers:
              25 Start watching this issue

                Created:
                Updated:
                Resolved: