DOS Alert in 5.3
The closed connection count increases after you call devices and permissions API.
The internal cache is refreshed.
Seems like we have opted for in-demand cache rather then redis cache.
You can try this with servers having 1000+ devices. 400+ Users. People with just maintenance use case and less devices may not be affected.
It’s not issue to have 1000+ devices with 400+ users.
The issue is that when the one user account has all devices.
When this happens it remove all cache for this user and reloads it.
Anton as per your below response : "That's expected. It invalidates everything linked to the user." https://www.traccar.org/forums/topic/dos-alert-in-53/page/2/#post-74918
So the cache reset and device disconnect experienced.
Gradually remove the cache scripts from the devices and permissions api works.
But now it is group api.
Cache implementation is to be checked in totality with servers in production.
I am personally interested in improving the time it takes to add a device. I can pitch in to get this feature implemented. If anyone else is willing to contribute, please email Anton. He already has my information.
We have updated cache invalidation, so the issue with slow API calls should be resolved now.
No, it doesn't. Where did you get this idea?