Fix Bad Randomly-High Latency Connections

The problem, (this is a common issue which been desribed aplenty here and on reddit): Some connections between client and server "randomly" have a high latency when many connections (to the same server location) are good. This then can often be "fixed" by reconnecting - like going through portal to hideout and coming back.

Unfortunately this "fix" has several obvious issues like losing a portal. In cases where there is only one portal this is particularly bad as in entirely useless.

There should be a proper automatic algorithm to deal with this. It would likely be easy to implement too. Even a manual "/reconnect" command or sth like that would be better than the current situation.

it isnt easy to implement

it is not the connection that is 'laggy' it is the instance that is dying due to one of many reasons

to 'reconnect' you need to recreate the same instance on a different, not dying node - this is easy, all you need is seed. and transfer the 'state' from old, already dying instance to this new one. this sounds easy but can be more than tricky in reality. then you need to get connected to this new instance - this also can be a challenge

given that the problem isnt widespread i doubt anything like that is going to happen

"

sidtherat wrote:

it is not the connection that is 'laggy' it is the instance that is dying due to one of many reasons

This topic is about laggy connections that can be fixed by leaving map and reentering it. So presumably the issue is routing. Bad instances are another thing, this topic is not about bad instances.

"

sidtherat wrote:

you need to recreate the same instance on a different, not dying node - this is easy, all you need is seed. and transfer the 'state' from old, already dying instance to this new one. this sounds easy but can be more than tricky in reality. then you need to get connected to this new instance - this also can be a challenge

If they have proper code they can do this kind of thing. But this is not what this topic is about. And they probably dont have proper code which means you are probably right in that that would be hard for them to do.

"

sidtherat wrote:

given that the problem isnt widespread i doubt anything like that is going to happen

I dont know how widespread the problem in this thread is but I remember reading about others having it too in various reddit posts, including non-EU. But it seems to be more affecting EU people. Presumably due to specific EU internet issues.