The workload pie gadget in Jira dashboard show wrong data about original estimate field. Please see the attached screen shots. The gadget shows 6h for this jb3 user but in JQL it shows more.
Couple of things.
If your concerned about privacy - you'll want to reach out directly to Atlassian Support.
https://support.atlassian.com/contact#/
Just an Atlassian Nerd
Hello @Jason Liu
Looking closely at your screenshots, the filter is set to assignee = sbiswas3 (Sumedha), but the 6h value belongs to jb3 (Jayakrishna). The gadget groups by Jira’s Assignee field, so please add AND assignee = jb3 to the exact saved filter used by the gadget.👀🧐 Then compare individual Original Estimate values, not the rolled-up total. The visible values for Sumedha add up to 28h, while the gadget shows 26h, but the screenshot doesn’t show the column headers. If the totals still differ after using the same filter and refreshing the gadget, it’s worth investigating further.
Cheers,
Arek🤠🧙♂️
Hi Jason,
One more thing I noticed in your screenshots. Your filter doesn’t select people by Assignee. It uses “Dev Assignee” and “Developer”, and jb3 is in both lists. But the gadget is set to Statistic Type “Assignee”.
So the two numbers answer different questions. The search tells you how much work has jb3 as Dev Assignee or Developer. The pie tells you how much of the filter’s work has jb3 in the Assignee field. If jb3 is the developer on a work item that is assigned to somebody else, those hours go to the other person’s slice.
A quick check: run the filter with AND assignee = jb3 added, like Arek suggested, and add up the Original Estimate column. I would expect 6h. Then run it with AND "Dev Assignee" = jb3 and I would expect your higher number.
AND assignee = jb3
AND "Dev Assignee" = jb3
If that’s it, you can change the gadget’s Statistic Type to Dev Assignee, if that field is offered in the list. Then the pie is grouped the same way as your filter.
Here's jb3 user filter result.
If we only count jb3 as developer and assignee tickets, it's not right, I think.
Thanks for the screenshot. My guess was wrong then. With assignee = jb3 the search shows a lot more than 6h, so the Dev Assignee field is not the reason.
The gadget only adds up the Original Estimate of the work items its saved filter returns. So if the numbers differ this much, I think the search and the gadget are not running the same query. In your screenshot the search page shows “Save as” and not the name of the filter, so it may be a changed copy.
Could you try this: open the saved filter “Overall - EMEA - Status not in Review” itself from Filters, without changing anything, add AND assignee = jb3 at the end and check the total. If that also gives 6h, compare the two queries line by line. If it gives the high number too, then it’s the gadget, and I would remove and add it again to rule out caching.
I think your screenshots show the answer. Look at which work items have an estimate. On the standard work items assigned to jb3 (the ones without a parent key in the summary) I count 2h + 2h + 2h = 6h, which is exactly what the pie shows. All the other hours are on sub-tasks.
So it looks like the gadget on your site counts only standard work items and leaves the sub-task estimates out.
You can test it quickly. Run your search with AND issuetype in standardIssueTypes() at the end and add up the Original Estimate. If you get 6h, that’s the explanation.
If your team estimates on sub-tasks, this gadget won’t show the real workload. The Two Dimensional Filter Statistics gadget won’t sum hours either, so for hours per person the Time Tracking report or an app like eazyBI is the usual way.
Hi Pavel, so the gadget only counts standard issue types estimates, right? Is there any configuration to change that or is it by default? Only easyBI can do sum with sub-tasks?
Hi @Jason Liu ,
Two steps: one to confirm the cause, one to get the real number.
1. Confirm it. Open the gadget's saved filter and run it twice:
<your filter> AND assignee = jb3 AND issuetype in standardIssueTypes()<your filter> AND assignee = jb3 AND issuetype in subTaskIssueTypes()
If the Original Estimate values in the first search add up to 6h, the pie is leaving out the sub-task estimates. I couldn't find this documented for the Workload Pie Chart gadget, so if it confirms, it's worth reporting to Atlassian support.
2. Get hours per person, sub-tasks included. Run the filter without the extra clauses, add the Original Estimate column (not Σ Original Estimate, which adds the sub-tasks' hours to the parent and would count them twice), export to CSV and pivot by Assignee in a spreadsheet.
If you need this refreshed regularly rather than as a one-off, an app can do the sum for you. I work on Advanced Export, so I'm biased, but it runs SQL over your Jira data ( https://marketplace.atlassian.com/apps/1217474 ):
SELECT u.displayName, COALESCE(SUM(CASE WHEN NOT t.subtask THEN w.timeoriginalestimate END), 0) / 3600.0 AS standard_hours, COALESCE(SUM(CASE WHEN t.subtask THEN w.timeoriginalestimate END), 0) / 3600.0 AS subtask_hours, COALESCE(SUM(w.timeoriginalestimate), 0) / 3600.0 AS total_hours FROM work_items w JOIN users u ON u.id = w.assignee JOIN issuetypes t ON t.id = w.issuetype GROUP BY u.displayName ORDER BY total_hours DESC;
standard_hours is what the pie seems to show; total_hours is the real workload. You don't have to write the SQL yourself: ask the app's Rovo agent for "original estimate hours per assignee, split by sub-tasks" and it writes the query and can schedule the report by email or upload to SFTP.
standard_hours
total_hours
Example query result on sample data
Documentation goes into detail on how to set it up and how to use Rovo.
Could you share what the two searches in step 1 add up to?
Hi, The quickest way to find where the hours go is to compare the two sets of issues directly. Open the gadget's settings, note the filter or project it uses, and run that exact filter in issue search. Then add both the "Original Estimate" and "Σ Original Estimate" columns. If your JQL total matches the Σ column but the gadget shows 6h, the gap is sub-task estimates being counted differently from parent estimates. If the issue lists are different, the gadget's filter is narrower than your JQL. Also make sure the gadget's time field is set to Original Estimate and not Current/Remaining Estimate. If it helps, we build https://marketplace.atlassian.com/apps/1230490/custom-dashboard-charts-and-gadgets?hosting=cloud&tab=overview , which can sum Original Estimate by assignee and always shows the JQL it runs.
I tested it to be sure, and I need to correct myself. On our Cloud site the Workload pie chart does count sub-tasks. I made a task with 2h and two sub-tasks with 3h and 4h, all assigned to me, and the pie showed 9h for me.
So it’s not general gadget behavior. You are on Data Center, which I can’t test, so two things to check:
Run your search with AND issuetype in standardIssueTypes(). If the total is exactly 6h, then on your version the gadget leaves sub-tasks out, and I would ask Atlassian support if that’s expected for your version.
If it’s not 6h, compare the gadget’s saved filter with the search you are running. Open the filter itself from Filters, not a changed copy, and check the total there.
There is no setting in the gadget for sub-tasks, as you saw in its configuration. If you need a quick number in the meantime, export the search to Excel and sum Original Estimate per assignee.
Hi @Jason Liu,
I did a quick test and it appears to work correctly to me. The gadget includes also the sub-tasks in the calculations.
In the latest versions of Jira Data Center, this gadget is provided by the Atlassian's Jira Charting Plugin. So please check if you have this plugin installed, and, if so, make sure it is running the latest version.
If you don't manage to resolve the issue, you could try apps from Atlassian Marketplace that provides a similar chart, and compare the results. If other apps show this correctly, you could contact Atlassian Support.
In case you want to try an app, our Great Gadgets app offers a Pivot Table & Pivot Chart gadget that can provides a similar chart.
To have such pie chart, the gadgets should be configured like this:
Hope this helps.
Danut.
Hi @Jason Liu, my name is Thiago, a support engineer from the Appfire Team.
Because of the AND logic connecting the user fields, this view is currently hiding:
As an alternative dashboard view, Rich Filters for Jira Dashboards can group issues by assignee and calculate the sum of the Original Estimate time-tracking field for each group. Its Rich Filter Statistics gadget also supports an additional working query, so you can explicitly control which issues are included, and it can display the results as a chart or table. These capabilities are described in The Rich Filter Statistics Gadget. You can find Rich Filters for Jira Dashboards on the Atlassian Marketplace here.
This would give you a separate calculation to compare with the native workload gadget, rather than changing the native gadget’s behavior.
Please contact our support if you have any other questions about this. Best regards, Appfire support team.
For standard issue type search for this user, it seems match the 6h result. Our Charting app version is 7.0.0. Our Jira is 11.3.12 Data Center. But I saw Danut's gadget counts sub-tasks. So if I want to count sub-tasks original estimate and not installing other apps. Is that possible?
Your filter contains this condition, which make it return only parent tasks, the sub-tasks being excluded.
Try removing this, or create a new filter without this condition and configure the gadget to use it. This should make it count sub-tasks as well.
If you want to count only sub-tasks, then change this condition into "AND issueType in subtaskIssueTypes()"
To answer your follow-up question on counting Sub-task Original Estimates natively without installing another app:
issuetype in standardIssueTypes()
Task
Story
Bug
2h
6h
timeoriginalestimate
If you ever want more control over how Original Estimate, Time Spent, or Remaining Estimate is broken down by Assignee and Issue Type on your Jira Data Center dashboard, you can also look at Easy Reports - Custom Charts (Free edition available) or Sum Up Reports for Jira (both available for Jira Data Center):
Dev Assignee
Developer
Hi Danut and Mehmet, the original problem of this post was that piechart was not matching the filter result. The standard issue type screen shot was added to check if the 6h estimate of user jb3 was standard issue type estimates. The original filter doesn't have it. But it still only shows standard issue type estimate sum-up. That's the question I wanted to ask.
Thank you for clarifying! That makes complete sense—I see now that AND assignee = jb3 AND issuetype in standardIssueTypes() at the bottom of your 6th screenshot was just an unsaved test (EDITED) to check whether jb3's 6h on the chart matched his 3 standard tasks (DXGWP-21577, DXGWP-16857, DXGWP-13551).
AND assignee = jb3 AND issuetype in standardIssueTypes()
EDITED
jb3
DXGWP-21577
DXGWP-16857
DXGWP-13551
Since you are on Jira Data Center 11.3.12 with jira-charting-plugin 7.0.0, inspecting the decompiled bytecode of WorkloadPieChart.class in jira-charting-plugin-7.0.0.jar reveals a calculation bug when a filter includes issues without an Original Estimate (Original Estimate = EMPTY):
jira-charting-plugin 7.0.0
WorkloadPieChart.class
jira-charting-plugin-7.0.0.jar
Original Estimate = EMPTY
In Jira 11 (jira-charting-plugin 7.0.0), WorkloadPieChart.calculateWorkloadDataset() was rewritten to aggregate via TwoDimensionalAggregator, grouping rows by distinct Original Estimate values (estimateSeconds) and columns by user.
WorkloadPieChart.calculateWorkloadDataset()
TwoDimensionalAggregator
Original Estimate
estimateSeconds
However, inside WorkloadPieChart.calculateWorkloadDataset(), when the loop encounters the null estimate bucket (representing issues in the filter where Original Estimate is empty), it executes:
null
if (estimateSeconds == null) { result.put(assignee, 0L); // Bug: overwrites the user's running sum with 0L instead of putIfAbsent! }
Because twoDimensionalStatistics.getRowToColumnValuesMap() iterates over an unordered HashMap, if a user has even one issue in the filter with an empty Original Estimate (in your screenshots, jb3 has 4 unestimated issues—DXGWP-21797, DXGWP-21793, DXGWP-20908, DXGWP-20285—and sbiswas3 has DXGWP-16734), hitting the null bucket mid-loop resets that user's running total back to 0L, wiping out every Original Estimate bucket that was summed before the null bucket in the map!
twoDimensionalStatistics.getRowToColumnValuesMap()
HashMap
DXGWP-21797
DXGWP-21793
DXGWP-20908
DXGWP-20285
sbiswas3
DXGWP-16734
0L
Add AND originalEstimate is not EMPTY to your saved filter Overall - EMEA - Status not in Review and save it.
AND originalEstimate is not EMPTY
Overall - EMEA - Status not in Review
That excludes the null estimate bucket from calculateWorkloadDataset() so it never resets a user's running total mid-calculation. (This does not appear to have a public JRASERVER ticket logged yet—if adding AND originalEstimate is not EMPTY fixes the totals, you can share WorkloadPieChart.calculateWorkloadDataset() with Atlassian Support so they can open an official bug report for jira-charting-plugin 7.0.0.)
calculateWorkloadDataset()
JRASERVER
Could you try adding AND originalEstimate is not EMPTY to your saved filter and let us know if the pie chart totals match?
It looks like you're new here. Sign in or register to get started.