Hi everyone,
I have a rather complex question regarding any potential performance issues with Grist. Let me start by describing my Grist project:
- self-hosted Grist on server
- accessed from Windows
- about 80 tables
- 62,000 rows, 10.8 MB data size, 0.76 GB attachmens according to the raw data page
- (in my view) quite complex access rules for almost every table
- document is made available to the public (although it can only be accessed within a VPN)
Here are the issue:
- We have set up Grist’s QR code widget to produce QR labels for our equipment. This link leads you to the row of the equipment via the linkkey property. The document is public so that people do not have to be logged in to see the content of the row. This works great whenever one is logged in. However, we already observe performance differences on our system: While Edge performs with acceptable speed, Firefox tends to be much slower. Does anyone know why?
- When I use a QR code in my browser, Firefox is unable to open the page and simply stalls. The same happens in Safari on my mobile phone. In Edge, it is also slow, but at some point, the intended row appears. We suspect that it might have anything to do with access rules because the problem seems to disappear when we remove large chunks of the access rules in a trial copy of our document. However, we do not understand why evaluating access rules would lose so much performance compared to a regularly logged in user. We also tried to put an access rule in that checks whether the user is logged, but it does not seem to help. Can anyone understand what might happen?
For now, I would not except an immediate solution, but rather a suggestion for what we should look at to find the issue.
Regards