The Label Studio cloud service provides scalability and low-latency for labeling user flow; however, there are some known limitations that could lead to performance degradations:
Local network provider latency
Sometimes page slowness can be affected by local network provider latency. Open your browser’s Network settings and check timings. Try again to refresh the page after some time.
High traffic
When there is high traffic to app.heartex.com , you may experience temporary page slowdowns. Typically it takes no longer than a few minutes to restore to the normal condition. Please be patient, as we’re working on enhancing predictability of scaling up the workflow in these cases. Please contact us in case you see some unusual slowdowns, by providing the exact time, project and user account
Data-related pages: Data Manager, Reports
Some of the pages’ performance depends on the volume of the data you store in the project. The bigger the project size in terms of the number of tasks, columns and created annotations, the slower are expected times to open Report pages (e.g. “Dashboard”, “Members”) or apply filters on the “Data Manager” page.
Try to reduce the number of the data in project. If the problem persists or if you see the page becomes non-responsive or result in error modals (e.g. Gateway timeout) - please contact us via form by specifying the time, project and user account.
Export snapshot takes a long time
There can be a problem with underlying job scheduler that may take a while to execute the export of jobs for your project. You can close the export dialog window, continue working and come back again to check the export status. If you see the error message - providing your project ID / URL.
Export snapshot takes endless time
Most likely RQ Workers are
- overloaded with other jobs
- setup incorrectly in your on-premise deployment
To inspect this issue you should open /django-rq page and see number of workers in Workers column:
If you see 0, it’s definitely a problem with your rq worker setup, you have to connect with your devops team and check what rqworker containers are running.
Also you can go inside of Failed jobs and check the error message in the each job, it will provide debug information to solve the issue.
Annotation result or Annotators filter is slow
Try upgrading to the most recent version of Label Studio Enterprise. This filter was significantly updated in Label Studio Enterprise 2.4.0 (and beyond).
Most of the API works slowly, the LS instance has more than million tasks
Try to run postgresql VACUUM, ANALYZE and REINDEX
Export conversion to other formats ends with Gateway Timeout error
As a workaround you can convert JSON manually on your local machine. Follow this link for instructions.