EVE General Discussion

hello, yes, me too.

its always my main that dc and my alt is just fine

Local Time: 19:41 Pacific Us - Socket Closed

It can be caused by a whole lot of different things. For instance, I used to get socketed out constantly when playing with two accounts after switching to a GPON network. I started playing with different settings at random, and it turned out that configuring DMZ for my computer's IP on the router fixed the problem. I'm no tech guy and can't explain why, but it worked.

Generally socket closed is a problem with the general internet, and nothing to do with CCP's servers at all.
Lrn2internet.

Who decides when the socket is dead and should be closed? Why doesn't the client know to ping the server once in a while to keep the socket open? This is the only game I've played where I've seen this error, not saying it doesn't happen in other games but the must have a better way to handover your connection so it doesn't drop completely.

Pleace create traceroute to TQ and file a bug report. Posting anything on this very forum wont solve any problems.

Really? FPS games are more tolerable of a high ping? Eves servers are some of the slowest(tick-rate) around, if your ping is less than 1000ms then you're not going to miss a tick. I don't believe that CCP are incapable of doing anything about this issue.

Also this... I live in the UK and when getting a spate of socket closed errors I use my VPN to connect to another server in London!

OK, sockets closed - the bane of every one's life when it happens. Socket closed happen when the game detects that the connection between you and TQ is unreliable - this help center article gives more details on the reasons for socket closed. Other programs may continue to work fine because their traffic may not be traversing a problematic route or because the application is more tolerant to network issues.

Usually a socket closed is a result of a networking blip between you and CCP's servers. This could be in your network, your ISP's network, any intermediate networks that your traffic passes through or even in CCP's network. When the latter occurs, it tends to disconnect lots of people at once and is picked up by their monitoring apps - this is part of the reason that the default GM response is "your ISP, not us". For quite a few of the people I have been helping with this recently, the issue has either been their local wifi, ISP issues or they have been affected by the number of broken sub-sea cables in the Australia/Asia/Pacific area.

So, how do you solve an ongoing socket closed problem? well the first step is to find out where the issue is and who is responsible for it. So if you are having an issue, please post where in the world you are, who your ISP is, what type of connection yours is and how you connect to your local router (wifi or wired).

An "easy" way to get some idea of where a problem may be is to grab a program called Pingplotter and set it going with a trace to client.eveonline.com while you play. When you get a socket closed, grab a screenshot of the pingplotter window and have a look at where the packetloss is and/or post it (I will gladly have a look at any posted pingplotters and help analyse them). Pingplotter is a good first step but it is notguaranteed to pick up every issue.

Another diagnostic is to play through a VPN. This can be a good short-term work around or a longer term solution on a really poor connection. If you don't get socket closed errors while connected to a VPN but do without it is indicative that there is a network problem somewhere. Using the VPN changes the route that your traffic takes to get to the cluster and the route that your traffic takes to the VPN server effectively becomes more reliable for the traffic you send up the VPN because of how they work. VPNs are not without their own issues though - they can increase your latency and decrease your bandwidth.

If pingplotter doesn't show anything and a couple of different VPNs don't help, then it is definitely worth reporting it to CCP along with all the diagnostic info you have gathered. If you get fobbed off with "it's your ISP" point them at the pingplotter showing no issues and escalate it.



The move to the new servers on the 29th seems to have caused some people's routing to the cluster to change. For some, this has been a more reliable route, for others a less reliable one. Also, on the 29th a large number of subsea cables near Singapore were damaged by seismic activity, causing some connectivity issues for people in that part of the world and causing their traffic to take different routes, adding load to infrastructure in other parts of the world.



Post a pingplotter when you get a disconnect. I am on VM and rarely have any issues however VM are known for overloading their local infrastructure at times and the Superhubs can have interesting wifi issues.

---

Finally, a word on analysing Pingplotters:

• 100% packet loss at a hop is not indicative of a problem if traffic continues past it - it just means that that hop does not respond with ICMP messages or to ping.
• A single hop with packet loss that does not propagate is not indicative of a problem. It just means that that hop is loaded and is proritising routing traffic over responding to tracert/ping.
• Packet loss from the start that propagates the length of the route is indicative of a local problem on your network.
• Packet loss that starts part way along a route and propagates is indicative of a problem along the route.
• Intermittent Packet loss that starts at client.eveonline.com is potentially indicative of an issue with the peering link between CCP and that transit provider. This is worth reporting to CCP as they may have some influence in getting them to sort it.