Since a while (unfortunately don’t remember the ha release, but its been a few months and ha updates since then) seemingly at random I have a spinning wheel displayed when looking at an entities history or logbook. It works fine for a while - all the data is displayed basically instantly. Except when it doesn’t. And its not bound to a specific entity. It can take up to minutes to show eventually the data, but with no significant cpu/disk/network load on my docker host.
I have a somehow complicated network setup so I assumed it has something to do with websockets getting terminated prematurely.
browser/companion app → haproxy → traefik → docker swarm → ha
So I’ve simplified my network setup as following:
- browser/companion app → haproxy → docker swarm → ha
- browser/companion app → docker swarm → ha
- browser/companion app → docker swarm in host mode → ha
And still I can reproduce the problem with the latest Safari/Chrome/Firefox/ios Companion App. Happens over the Internet and in my LAN.
There are no relevant log entries in ha. The only errors I could find which might be relevant show up in the Browsers Developer Console.
I’ve googled, tcpdumped and debugged an unreasonable time. Down every rabbit hole I could find. Thats why I decided to post today. Anybody seeing this too on their installation? I can’t be the only one. 
Its been 6 days since the change (with a purge_keep_days of 7), so I think its fair to say that this did it.