Status of Game? It sucks!
-
Who would have thought that out of all things this would be the downfall of dual gap lol
-
We don't know what is causing these problems. We identified some reasons and resolved then (e.g. ddos protection of the main server, solving coturn cpu issues, potential network stress caused by a game change).
There is no visible reason why the problem still exists and we also received feedback that it is fixed for some of you. We have no tools to diagnose these problems any further, we have to develop them first.
We cannot rollback FAF as the system is not designed for rollback. Also there is no guarantee that a rollback actually fixed anything.
FAF had always been plagued by connection issues, and the world around is still changing. Back then we faced 24h disconnects, today we have providers pushing people into improper ipv4 ip pools with natting on provider level, data centers changing their connectivity and firewalls, we have more aggressive routers and security software than ever before. These problems doesn't even need to come from us, it could be imposed externally.
What we need is reliable data. Unfortunately available developer time in that area is more spare than it used to be a few years ago.
-
There is no roadmap because at the moment we don't even know for sure what the problem is. We need more telemetry to diagnose the problem, but this has to be coded first. Recently some people have come forward that wanted to help with this but I don't know the latest status.
Edit: Brutus was two minutes faster
-
Is it possible to identify the issue occurring in server logs? Then work out which date it started. Then look at changes around that date.. spinning up coturn servers or whatever
-
No. It's peer to peer connections that break. So it's not even touching the main server (anymore). Coturn servers are dedicated machines now. So assuming it would be an occasional coturn problem it should only affect players using the relay.
And these kind of network scenarios can't even be manually tested. -
When users encounter such unusual connection issues, what kind of information could they provide to help?
OS, network card, security software, router, internet provider, or/and any tracert results from CMD?
-
We would need all that info from every player in the game. Which is just infeasible to sort through manually. Thus the need for more developed telemetry
-
I would have thought when games send the score result back to base you would get some idea of number of players still connected at the end, or how many dropped out, or something, anything to measure connectivity.
Failing that maybe use something less direct like avg time of ladder match. If connections are really super broken you might see a drop there. Yes I'm clutching at straws
How do we even know if the problem really exists or not
-
There is no message about why a player disconnected currently in any of the messages to the server. So just seeing who is connected is not good enough.
-
I wonder if something similar is possible to the CPU test. Just sending one minute of data to everyone in the lobby. A proper stress test that shows how likely the game will run smoothly.
-
In an ideal world nobody would even need to use the coturn servers. Everybody would just be able to establish direct connections. But obviously they can't. So now there is a whole range of possible issues why a connection could not be established or drops. Troubleshooting that is quite hard
-
Having the number of disconnects is a huge clue. Plot it over time. See the date it started. Look at deployments near that date.
-
@nine2 said in Status of Game? It sucks!:
How do we even know if the problem really exists or not
It has always existed. There were always scenarios where player changed something and were never able to play. Nothing we can do about. Big question is: How do we know that things got worse? When do we know that it is good enough now and stop trying to fix the impossible?
-
We can't plot the number of disconnects over time because we can't measure those at the moment. That's why we first need to develop better tooling as explained
-
Makes sense.
-
the ingame disconnects got a lot better. like 4 weeks ago no match worked without disconnects and multiple con issues between multiple players.
now its the problem that many players cant connect to lobbies, but here again i got the feeling that its better than a week ago. -
There are two problems. One is the instability and the other that ICE doesn't reconnect the players - a feature that worked once upon a time. Wasn't the reconnection feature even the number one goal why ICE was invented?
And I don't know if it is only me, but when I quit FAF, it stays most of the time in the task manager and I have to end the task.
-
No the reconnection feature was a nice side effect, the main problem it solved was establishing connections without port forwarding (similar to what Skype made so successful)
-
@brutus5000 said in Status of Game? It sucks!:
No the reconnection feature was a nice side effect, the main problem it solved was establishing connections without port forwarding (similar to what Skype made so successful)
But why doesn't the side feature work anymore? That shouldn't have anything to do with server load, ddos, broken hard drives and so on. Are there no logs, client or server that exactly causes the impossibility to reconnect?
-
J Jip referenced this topic on
-
@closed said in Status of Game? It sucks!:
And I don't know if it is only me, but when I quit FAF, it stays most of the time in the task manager and I have to end the task.
Maybe this is worth mentioning, but the problems my (closed) playgroup has had seems to also involve the ICE adapters, and people not being able to quit the game properly without a black screen and involving Task Manager.
We play tonight, so I guess we have to cross our fingers that some reinstalls we did help.
-
A Arondy referenced this topic on
