-
Type:
Bug
-
Status: Closed
-
Resolution: Completed
-
Affects Version/s: 7.2.X, Master
-
Fix Version/s: 7.2.X, 7.3.10 DXP FP1, 7.3.10.1 DXP SP1, 7.4.13 DXP GA1, Master
-
Component/s: Content Dashboard
-
Branch Version/s:7.2.x
-
Backported to Branch:Committed
-
Fix Priority:4
-
Epic Link:
-
Sprint:TANGO | #47 DEV | Oct09-Oct23
-
Git Pull Request:
Description
We have executed the SignIn performance test with 100 segments, and the result shows a very important regression, you can check the full result in the table below:
Portal Version | Git Id | Session Count | Mean Time of Login | Portal CPU Usage (%) | DB CPU Usage (%) | Session count changes(%) |
---|---|---|---|---|---|---|
BaseLine | e1efdfd9 | 39050 | 262 | 96,96,96 | 0,0,0 | |
Segment(100 segments) | e1efdfd9 | 1936 | 34200 | 60,60,60 | 28,28,28 | -95.04225352 |
You can find the full results here
Also, you can download the profile from here.
Proposed solution
Taking a look into the profile, there is a query that takes most of the time
So using -1 as the groupId in the Search Container will avoid calling the groups query inside com.liferay.portal.search.internal.spi.model.query.contributor.GroupIdQueryPreFilterContributor
- is a dependency of
-
LPS-120006 Performace issues with segments due to excesive Elasticsearch queries
- Closed
- is related to
-
LPS-110846 Many Inactive Groups (Sites or deactivated Users) causes User Groups and Users and Organizations pages to load slowly or fail to load
- Closed
-
LPS-121537 GroupLocalServiceImpl.getGroupIds always retrieves results from database
- Closed