It happens often that certain chatmail relays are down at random times of the day. Sometimes, scheduled maintenance is announced 5-6 minutes before the impact.
Could we perhaps document guidance about best practices for operators such as announcing scheduled maintenance a day ahead, or at least within a few hours, so users had time to prepare (or possibly switch relays in time)?
a PR or a forum post with documentation about best practises in scheduling maintenance is probably very welcome, just go ahead!
current focus is an alternative stance, that errors and timeouts are okay. there is very concrete work on that:
with “multi relay”, when fully rolled out, it will be fine if a relay is not working for some time or even forever. details needs to be figured out and are discussed a lot (see github), but as a rule of thumb, we want things being as automatic as possible.
user onboard on multiple relays
user receives messages on all relays - and chat partners will be aware of all relays
when a relay goes down, the sender uses the other ones, as well as the receiver listens on all of them
this will give piece of mind for both, user and relay ops.
“right to exist” on all sides.
current state is that your identity already is no longer your address, but exist only on your device. and you can already manually add/remove relays and listen on all of them.