Uploaded image for project: 'Jira Service Management Data Center'
  1. Jira Service Management Data Center
  2. JSDSERVER-6047

Workload report loads very slowly

XMLWordPrintable

      Summary

      The workload reports page shows list of agents, their assigned issue counts and number of breached and ongoing issues.
      To find the breached and ongoing issues we completely calculate everything to do with SLAs leading to slow performance.
      The configuration in the attached screenshot took 23 seconds to load.

      Steps to Reproduce

      1. Try to load a Workload report

      Expected Results

      Report loads in a reasonable time

      Actual Results

      Report take a very long time to load like ~30secs

      Notes

      For all issues that have an assignee, we loop through all SLAs, and get goalSummaryData.
      In getting goal data, we retrieve everything about ongoing goals, including expected breach date that gets active ranges in nested loops. This is already O(n^4). It's possible that there is a worse flow.

        1. Screen Shot 2018-09-28 at 5.22.07 PM.png
          Screen Shot 2018-09-28 at 5.22.07 PM.png
          112 kB
        2. Screen Shot 2020-04-20 at 4.05.41 PM.png
          Screen Shot 2020-04-20 at 4.05.41 PM.png
          49 kB
        3. Screen Shot 2020-04-28 at 1.45.56 PM.png
          Screen Shot 2020-04-28 at 1.45.56 PM.png
          171 kB

            [JSDSERVER-6047] Workload report loads very slowly

            Jessie Chia added a comment -

            Bump. Users are not happy that they have to wait more than half a minute for the report.

            Jessie Chia added a comment - Bump. Users are not happy that they have to wait more than half a minute for the report.

            Jessie Chia added a comment -

            Still happening in 8.20.10.

            Jessie Chia added a comment - Still happening in 8.20.10.

            Denise Unterwurzacher [Atlassian] (Inactive) added a comment - - edited

            On a staging instance with no other traffic I loaded 2x Workload reports. Both returned a 500 in the browser after 60s. However, the backend process continued and impacted the CPU and file system read/write IOPS (write more than read, interestingly).

            As you can see, this impact lasted just under 1.5 hours.

            This can cause outages. Updating Symptom Severity to Critical.

            Denise Unterwurzacher [Atlassian] (Inactive) added a comment - - edited On a staging instance with no other traffic I loaded 2x Workload reports. Both returned a 500 in the browser after 60s. However, the backend process continued and impacted the CPU and file system read/write IOPS (write more than read, interestingly). As you can see, this impact lasted just under 1.5 hours. This can cause outages. Updating Symptom Severity to Critical.

            Denise Unterwurzacher [Atlassian] (Inactive) added a comment -

            This issue is still occurring on the Jira 8.5.x so I am reopening this issue.

            Denise Unterwurzacher [Atlassian] (Inactive) added a comment - This issue is still occurring on the Jira 8.5.x so I am reopening this issue.

            Bartosz Ornatowski added a comment -

            This issue has been fixed in 4.0 release and has been backported to Enterprise Release 3.16.2+

            Bartosz Ornatowski added a comment - This issue has been fixed in 4.0 release and has been backported to Enterprise Release 3.16.2+

            Graziano Liberati added a comment -

            Hi, any news on this bug?

            Graziano Liberati added a comment - Hi, any news on this bug?

            Ruchi Tandon added a comment -

            Thanks Jimbo.

            Our development teams are now aware of this bug and I hope they will be able to fix it soon .

            Ruchi Tandon added a comment - Thanks Jimbo. Our development teams are now aware of this bug and I hope they will be able to fix it soon .

            James H added a comment - - edited

            Hi Ruchi, we see the same thing - on ours it takes around 10 seconds to load from a project of < 10,000 issues on a Redhat server with 6 CPUs and 16GB RAM and Java allocation of 4GB.  Our database server is PostGreSQL with 6 CPUs and 8GB RAM.  Opening this screen is pretty much the only time the CPU makes a mark on the performance monitoring chart of the DB server:

            James H added a comment - - edited Hi Ruchi, we see the same thing - on ours it takes around 10 seconds to load from a project of < 10,000 issues on a Redhat server with 6 CPUs and 16GB RAM and Java allocation of 4GB.  Our database server is PostGreSQL with 6 CPUs and 8GB RAM.  Opening this screen is pretty much the only time the CPU makes a mark on the performance monitoring chart of the DB server:

              bfa5bb10b6a8 Arslan Qamar (Inactive)
              rtandon@atlassian.com Ruchi Tandon
              Affected customers:
              8 This affects my team
              Watchers:
              19 Start watching this issue

                Created:
                Updated:
                Resolved: