
- Subscribe to RSS Feed
- Mark as New
- Mark as Read
- Bookmark
- Subscribe
- Printer Friendly Page
- Report Inappropriate Content
on
01-30-2025
12:24 PM
- edited on
02-12-2025
05:21 AM
by
Oleks_Burkov
Steps to do before contacting support regarding performance issue
The article is about issues connected to situations when dashboards are loading lower than expected. Widgets are failing to load at all. What you should do and what information about errors to gather before contacting the support team to ensure faster resolution of the issue?
Which information to add when creating a ticket
- Record the .har file when loading the affected dashboard and one of the affected widgets. If many widgets are a choose one, which you will use as an example in all other steps (see here how to record and download a .har file)
- Disable all filters on a dashboard and load it again, while recording yet another .har file
- If you are using any widget type other than pivot, skip this step.
If you are using pivot, make a copy of the affected widget and change the type to ‘table’. Reload the widget, while recording .har file again
- In the .har file look for ‘jaql’ request. Copy the X-Request ID of any of the requests. You will need it later. Look through ‘jaq requests, do any of them have a red color and status code other than 200? Also note this information (see here how to find ‘jaql’ request and locate the X-Request ID and status code)
- Open the cube which is a data source for the dashboard
[Alt text: A screenshot of a data analytics interface labeled "Sample Healthcare." It shows a drop-down menu with an option to "Open Data Source." Below the menu, there is a table presenting years of admission data from 2012 to 2013. The date shown at the top right corner is April 22, 2021.]
6. Preview the table which includes the data used in one of the affected widgets. Can you preview the data from the cube? Specify this information in the ticket.
- If the data source for the affected dashboard is Elasticube and any of ‘jaql’ requests you found in step 4 has status code 50 or any of the failed widgets have the error ‘Safe mode’ when loaded, do the following:
7.1. Go back to the cube, which is the data source for the affected dashboard, and create a duplicated cub
[ALT Text: A screenshot of a software interface displaying a menu related to "Relationships" and "Build." The menu options include "Duplicate Model," "Duplicate As Live," "Duplicate with Direct Relations," "Export," and "Fiscal Year start." The interface features colorful nodes connected by lines, indicating a visual representation of data relationships.]
[ALT Text: A user interface window titled "Build" with options related to importing data in a sample healthcare dashboard. It includes settings for "Sample Data" with a field to limit the number of imported rows per table, currently set to 100. There are buttons labeled "Build" and "Cancel" at the bottom of the window.]
[ALT text: A dropdown menu titled "Change Sample Healthcare" displaying options for data sources. Options include "Sample ECommerce," "Sample Healthcare," "Sample Healthcare Performance Test" (highlighted), "Sample Lead Generation," and "Sample Healthcare Park." The date on the image is April 22, 2021.]
Are all widgets loading normally? Are they failing? Or still, loading longer than expected? Include this information in a ticket
- Check the Data group settings.
[ALT Text: A screenshot of a web application interface, specifically the 'Admin' section of the software. It displays a table with columns labeled 'Data Group', 'Build Node', 'Entity Node', and 'Exact Node'. There are options for 'Data Management' and 'Edit Data Group' highlighted, with the main content area showing a list of data groups, including 'Default Defaults' and 'Sample Lead Owner'. The layout appears simple and user-friendly.]
[ALT text: A screenshot of a user interface titled "Edit Data Group." It displays fields for "Group Name," "Build Node," "Query Nodes," and "Instances," along with options for "ElasticCubes" selection. There is a toggle switch labeled "Stop when idle" and a setting for "Idle Time (Minutes)" set to 30.]
- Check memory limits of query pod in Configuration
[ALT text: A user interface screenshot of the Sisense application showing the "Admin" section. It includes tabs for "System Management," "Configuration," and "Networking," with a focus on "System Management." The left sidebar features options like "User Management," "App Configuration," and "Server & Hardware." Various fields and buttons are present, indicating a management dashboard. The image appears slightly blurred.]
10. Run the following command from the Sisense server (Only for on-premise customers): kubectl -n sisense get deployments.apps query -o yaml | grep -A2 limits
11. Check Grafana to see how much memory is consumed. Here is the link to an article on how to check specific PO consumption in Grafana: Check these pods:
query
ec-qry (affected cube and all cube
- Login to the Sisense server and download the log files specified below. Make sure to proceed with this step right of performing all steps above (Only for on-premise customers).
query.log
combined.log
Before downloading the necessary logs, make sure that the X-Request ID from Step 4 is present in the logs. You may check how to identify if necessary X-Request ID if present in the logs.
You may check how to identify if necessary X-Request ID if present in the logs.
You may check how to download and identify necessary logs here.
Please raise a ticket with the Sisense support team and add all gathered information to the ticket.