Sudden strange behaviour from VW

1

I’ve had my selfhosted VW server running for over a year, with no problems of any kind.
Currently I’m on version 1.33.2

I’ve noticed that from time to time when I need to take down the server for maintenance, my browsers both in Windows as well as on my SmartPhone continues to correctly provide username and password credentials. I assumed this to be a result of local caching (?). How else could they continue to work when the server is shut down?
There is probably some mechanisms in place between the web client and the server that kicks in whenever synchronization is required. In between synchronizations, it most likely relies on the local cache, right?

But then to my point; Lately I have occasionally been asked by the web client to log in to the VW server using the master password. Not necessarily surprising, but when this happens (not often) I get stuck in a loop whereby this authentication request comes up again and again in a never ending round robin. No error messages or any other messages. It just keeps asking for the master password, but seems to ignore my entries.
If I check the server itself by logging on to the VW web UI, everything works ok. So obviously the server is running and the master password is correctly recognized.
If I visit a new www site for the first time, The VW web client comes up and asks me if I want it to save the new username/password combination for that site, and if I acknowledge, it does exactly that and by logging on directly to the server I can verify that the new credentials has been saved. So the communication between the server end the web client obviously works as it should.

What I’m curious about, is if anyone else has experienced the looping request for master password to authenticate?

2

I had observed this behaviour at quite an early stage of use. What I most often traced it down to, was the fact that I had either turned off or misconfigured routing to the server.

The local browser was then acting on a cache of earlier connections, which was clearly going to fail.

The give away was, that swapping to another browser could often resolve it, or a stop restart of the client machine, and lastly, working in Incognito mode removed the issue - i.e. cache was refreshed correctly to a correctly configured server, so it ‘worked’ or it ‘failed’.

I don’t encounter this at all now, as the server is a permanent and stable installation.