Evolution still tries to refresh emails while a WAN outage is occuring and Network Manager reports the Internet being inaccessible
At my parents' home, they have an unreliable Internet connection that randomly dies multiple times per day, even though the router is fine, so the LAN is up but the WAN is dead. Sometimes this happens at the same time Evolution wanted to load and display a message or to refresh a bunch of things.
On Evolution 3.50.4 from Fedora 39, as well as previous versions, Evolution will waste time and get stuck trying to do network operations (and spam the user with error infobars) even when Network Manager knows the WAN is down (as indicated in the system indicator, circled in red in the screenshot below):
In that screenshot, you can also see the spurious error infobars, and the bunch of cancellable operations accumulating in the statusbar at the bottom.
In my Evolution network settings, it has no proxy set and it is configured to use the default method (networkmanager) to detect online state.
When NM was previously online and connected to WAN and suddenly loses access to WAN (especially once it has experienced at least one socket timeout error), ideally Evolution should then understand that the Internet connection is having problems and should go into offline mode instead of trying to do impossible network operations, until NM says that full Internet connectivity is back online.