These forums have been archived and are now read-only.

The new forums are live and can be found at https://forums.eveonline.com/

Issues, Workarounds & Localization

 
  • Topic is locked indefinitely.
 

Socket Closed - Launcher only

First post First post
Author
Previous Topic Next Topic
Helios Anduath
The Scope
Gallente Federation
#181 - 2016-03-07 11:53:56 UTC
Quinn Jovakko wrote:
I hope this is allowed, but I recorded until I actually lost connection to eve. Here's the pingplotter file.

https://drive.google.com/file/d/0By8l9q8ACl3oN2N1aUd5OUZ0dEE/view?usp=sharing


ouch, at what point did the disconnect happen? was it the red block of packet loss at the end?

That capture is showing that there is definitely an issue (maybe not the only one) on your local network as you have packet loss between you and your own router that propagates the entire length of the trace. How do you connect to your home network - wifi or wired?

The first step to working out if there are any other issues is to "fix" whatever is causing packet loss on your own network.
Helios Anduath
The Scope
Gallente Federation
#182 - 2016-03-07 11:57:10 UTC  |  Edited by: Helios Anduath
Panther2707 wrote:
Been getting Socket closes every 5-15 minutes in eve, if im lucky i make it to 20 minutes before disconnecting (with and without using launcher).

http://imgur.com/3jlFGCL

Picture refers to my pingplotter, took the screenshot when i disconnected. Haven't been able to play for the last week with the disconnections.


The minor packet loss on a few hops is indicative that there could be an issue in Telstra's network. Given the sub-sea cable issues around Australia and the Asia-Pacific region at the moment, this is not entirely unexpected. EDIT: just to add that a lot of the cable issues started on the 29th so connectivity issues in the last week does tie into this.

Have you tried playing through a VPN? Some people find that it can help to stabilise things on marginal connections at an increase in latency. Testing through a VPN can also help to confirm/disprove whether it is connectivity issues along the path your traffic takes or something else.
Ravcharas
Infinite Point
Pandemic Horde
#183 - 2016-03-07 15:22:01 UTC
Helios, you seem to be on the case here. Could you spare a minute and explain why someone would get socket closed on just one account and not on any other? Always the same account, doesn't matter if it's the only one logged in or not.
Helios Anduath
The Scope
Gallente Federation
#184 - 2016-03-07 15:31:57 UTC
Ravcharas wrote:
Helios, you seem to be on the case here. Could you spare a minute and explain why someone would get socket closed on just one account and not on any other? Always the same account, doesn't matter if it's the only one logged in or not.


That I can't definitively say but if we can discount any networking issues, then it is something that CCP would need to look into and won't be able to fob people off with "it's your ISP".

When running multiple clients it is quite possible to get socket closed on one while another stays logged in as different activities have different traffic patterns. For example, a client that is undocked and active probably sends more traffic than a client that is docked without focus. A network blip could cause the active client to DC but not the inactive client.

Are you only getting the DCs when you do one particular activity that you do on the account with issues? is it just one character on the acount with problems? what does a pingplotter capture show during a disconnect on the problem account?
Elijah Sool
Kessel Run Industries
#185 - 2016-03-07 20:51:39 UTC
I'm a xs4all user from the netherlands.. and switching to a dutch address in a vpn solved my packet loss (which caused lag (small and severe) and disconnects :)
Helios Anduath
The Scope
Gallente Federation
#186 - 2016-03-07 21:22:17 UTC
Elijah Sool wrote:
I'm a xs4all user from the netherlands.. and switching to a dutch address in a vpn solved my packet loss (which caused lag (small and severe) and disconnects :)


As I have said in the thread before, that indicates an issue along the route that your traffic takes without the VPN. Please can you post a screenshot of a pingplotter capture around the time of a disconnect event. A tracert from when you are connected to the VPN would also be useful to see what route works.
Ravcharas
Infinite Point
Pandemic Horde
#187 - 2016-03-07 22:53:45 UTC  |  Edited by: Ravcharas
Helios Anduath wrote:
Ravcharas wrote:
Helios, you seem to be on the case here. Could you spare a minute and explain why someone would get socket closed on just one account and not on any other? Always the same account, doesn't matter if it's the only one logged in or not.


That I can't definitively say but if we can discount any networking issues, then it is something that CCP would need to look into and won't be able to fob people off with "it's your ISP".

When running multiple clients it is quite possible to get socket closed on one while another stays logged in as different activities have different traffic patterns. For example, a client that is undocked and active probably sends more traffic than a client that is docked without focus. A network blip could cause the active client to DC but not the inactive client.

Are you only getting the DCs when you do one particular activity that you do on the account with issues? is it just one character on the acount with problems? what does a pingplotter capture show during a disconnect on the problem account?


It's not just one character. I haven't pingplotted this go around. I remember I had issues before without the pingplot or tracert showing any obvious problems. It went away after a patch. It seems to happen when just idling in a station. So at first I thought it was some hardware along the way that caused it, but the other clients are fine - no matter how idle they are. So yeah, I'm at a loss. A packet loss, as it were.
Helios Anduath
The Scope
Gallente Federation
#188 - 2016-03-07 23:01:08 UTC
Ravcharas wrote:
It's not just one character. I haven't pingplotted this go around. I remember I had issues before without the pingplot or tracert showing any obvious problems. It went away after a patch. It seems to happen when just idling in a station. So at first I thought it was some hardware along the way that caused it, but the other clients are fine - no matter how idle they are. So yeah, I'm at a loss. A packet loss, as it were.


That one is definitely worth a ticket provided the pingplotter doesn't show anything. When making it, make it clear that it happens to one specific account while idling and include links to pingplotter screenshots annotated with disconnect times. When they respond with the "It's your ISP" line, escalate it.

One thing to check before making the ticket though is whether using a VPN helps or not. More diagnostics as it were.
Rhys Redin
The Scope
Gallente Federation
#189 - 2016-03-08 00:53:07 UTC
Helios Anduath wrote:
Rhys Redin wrote:
Getting multiple socket closures per hour when running multiple clients. What is being done to fix this?


There are lots of people who are running multiple clients with no issue so you are going to have to give CCP some help working out why you have issues.

Who is your ISP and where in the world are you?

Can you get Pingplotter running (link further up in the thread) and post a screenshot of what it shows when you disconnect on your client(s)? From that we should be able to get an idea of if there is an issue with your ISP, your local network or something further up the chain.


I don't pay subscription money to do CCP's job for them. Everything was fine until the data center migration. The problem isn't on my end.
Helios Anduath
The Scope
Gallente Federation
#190 - 2016-03-08 02:00:17 UTC  |  Edited by: Helios Anduath
Rhys Redin wrote:
I don't pay subscription money to do CCP's job for them. Everything was fine until the data center migration. The problem isn't on my end.


CCP can't diagnose every issue without user involvement, especially when there are innumerable hardware, software and networking configurations out there. This is how support is in the tech field and if you actually check any support contracts that you pay for, you will normally find terms that require your co-operation in diagnosing issues. If you are not prepared to do your bit in helping to diagnose the issue, they can't help you and it shows an attitude of over-entitlement and gives the impression that you don't want to get the issue fixed.

You pay CCP to allow you un-guaranteed access to a service, not to diagnose problems that are usually to do with things outside of their control. With socket closed errors, a lot of the time it is due to a problem outside of their control. Their network monitoring tends to pick up any issues with the servers but it can't diagnose problems with your setup, your ISP's setup or any of the other intermediate networks that your traffic passes through.

Now, you say it started about the time of the data centre move and new servers - are you by any chance located in Australia, New Zealand, Singapore or any other countries in that area? If so, there was a seismic event on the 29th that severed several undersea cables causing disruption to traffic in that part of the world. This is on top of the cable issue affecting Australia from early Feb.

EDIT: The other thing is that the data centre move may have resulted in a change of route that your traffic takes across the internet to get to CCP. It is possible that there is an issue with some routes but again, it needs input from you to help determine.

I could go on about how attitudes like the one you have displayed mean that if there is an underlying issue, it won't get fixed as people won't help themselves but I doubt you would take it onboard so by all means just keep sitting there feeling over-entitled.
Panther2707
Exit-Strategy
Minmatar Fleet Alliance
#191 - 2016-03-08 06:42:20 UTC
Helios Anduath wrote:
Panther2707 wrote:
Been getting Socket closes every 5-15 minutes in eve, if im lucky i make it to 20 minutes before disconnecting (with and without using launcher).

http://imgur.com/3jlFGCL

Picture refers to my pingplotter, took the screenshot when i disconnected. Haven't been able to play for the last week with the disconnections.


The minor packet loss on a few hops is indicative that there could be an issue in Telstra's network. Given the sub-sea cable issues around Australia and the Asia-Pacific region at the moment, this is not entirely unexpected. EDIT: just to add that a lot of the cable issues started on the 29th so connectivity issues in the last week does tie into this.

Have you tried playing through a VPN? Some people find that it can help to stabilise things on marginal connections at an increase in latency. Testing through a VPN can also help to confirm/disprove whether it is connectivity issues along the path your traffic takes or something else.


Just tried running through a friends VPN, 45 minutes connected to eve so far no issue (17 minutes was my record for this week), will leave it running a bit longer but its almost definitely what you mentioned, the undersea cables since my issue did start at the same time.
Helios Anduath
The Scope
Gallente Federation
#192 - 2016-03-08 10:59:11 UTC
Panther2707 wrote:
Just tried running through a friends VPN, 45 minutes connected to eve so far no issue (17 minutes was my record for this week), will leave it running a bit longer but its almost definitely what you mentioned, the undersea cables since my issue did start at the same time.


hopefully it is the issue and everything will evenout when the cables are repaired (due later on in March according to news reports). Keep an eye on it and post back if you get any disconnects while connected to the VPN?

The VPN helps to stabilise things for two reasons: firstly, it potentially lets your traffic take a different route, avoiding any issues. Secondly, it uses a more reliable transport so packetloss is compensated for. These both obviously come at the cost of increased latency to your intended destination and reduced bandwidth so may not be a good long-term option.

Out of curiosity, I am wondering what route you take while you are connected to the van and what your route to the VPN server is? Any chance of a tracert of both?
Shade Millith
Tactical Farmers.
Pandemic Horde
#193 - 2016-03-09 01:27:36 UTC  |  Edited by: Shade Millith
I've been keeping track of which client is getting socketcloses, and I've noticed that it's always a client that's been idle for a while. Never had I had a socket close on a client that's been actively used.

If I'm running two clients, using one to keep an eye on another system, and one I'm actively playing on, it will always be the scouting client that gets a socket close. Same with three-five clients, the ones that are being used will never socketclose. Just the ones that are sitting idle.

I've also noticed that clients that have been sitting for a while can become extremely unresponsive for 10-15 seconds, unable to switch ship, activate modules, right click on things, etc. After a little time, the client fixes itself and continues on without issue.

I've also been watching PingPlotter, and while some socketcloses have a spike in PL, there are plenty that don't show anything.

This, combined with how it started when they did the server upgrade, is making me thinks that perhaps CCP changed something with how the server handles clients that haven't made any actions for a while, and it's having hiccups.
Helios Anduath
The Scope
Gallente Federation
#194 - 2016-03-09 02:34:58 UTC
Shade Millith wrote:
If I'm running two clients, using one to keep an eye on another system, and one I'm actively playing on, it will always be the scouting client that gets a socket close. Same with three-five clients, the ones that are being used will never socketclose. Just the ones that are sitting idle.

I've also noticed that clients that have been sitting for a while can become extremely unresponsive for 10-15 seconds, unable to switch ship, activate modules, right click on things, etc. After a little time, the client fixes itself and continues on without issue.


That is probably worth a bug report with as much info as possible once you have done one more test (see below). Is there any corelation between time idle and frequency of socket closures?

Shade Millith wrote:
I've also been watching PingPlotter, and while some socketcloses have a spike in PL, there are plenty that don't show anything.


While Pingplotter can be useful in finding packet loss, it won't always detect a problem. Routers can treat ICMP traffic differently or an issue may manifest between checks. This is why I have been suggesting people trying through a VPN as well.

Do you still get the idle-client disconnects when playing through a VPN?
Panther2707
Exit-Strategy
Minmatar Fleet Alliance
#195 - 2016-03-09 05:43:49 UTC
Helios Anduath wrote:
Panther2707 wrote:
Just tried running through a friends VPN, 45 minutes connected to eve so far no issue (17 minutes was my record for this week), will leave it running a bit longer but its almost definitely what you mentioned, the undersea cables since my issue did start at the same time.


hopefully it is the issue and everything will evenout when the cables are repaired (due later on in March according to news reports). Keep an eye on it and post back if you get any disconnects while connected to the VPN?

The VPN helps to stabilise things for two reasons: firstly, it potentially lets your traffic take a different route, avoiding any issues. Secondly, it uses a more reliable transport so packetloss is compensated for. These both obviously come at the cost of increased latency to your intended destination and reduced bandwidth so may not be a good long-term option.

Out of curiosity, I am wondering what route you take while you are connected to the van and what your route to the VPN server is? Any chance of a tracert of both?


Using a sydney server for VPN, first log is the route from me to the VPN, second log is from the VPN to the server. No issues at all last night, probably added another ~100ms to my ping but being australian im already used to high ping so its manageable until the issue clears.


  1    <1 ms    <1 ms    <1 ms  192.168.0.1
  2    20 ms    20 ms    20 ms  172.18.209.13
  3    19 ms    20 ms    19 ms  172.18.65.213
  4    20 ms    20 ms    20 ms  bundle-ether4.ken-edge901.sydney.telstra.net [203.50.12.104]
  
  5    21 ms    20 ms    21 ms  bundle-ether13.ken-core10.sydney.telstra.net [203.50.11.94]
  
  6    21 ms    20 ms    20 ms  bundle-ether1.ken-edge902.sydney.telstra.net [203.50.11.97]
  
  7    20 ms    19 ms    19 ms  sof1522705.lnk.telstra.net [139.130.197.166]
  8    20 ms    21 ms    20 ms  ae5.dar02.syd01.networklayer.com [50.97.19.151]

  9    19 ms    19 ms    20 ms  po2.fcr01a.syd01.networklayer.com [168.1.18.133]

 10    20 ms    20 ms    20 ms  5f.17.01a8.ip4.static.sl-reverse.com [168.1.23.95]



  1    20 ms    20 ms    19 ms  10.129.1.1
  2    21 ms    21 ms    21 ms  2.06.01a8.ip4.static.sl-reverse.com [168.1.6.2]

  3    20 ms    20 ms    20 ms  82.12.01a8.ip4.static.sl-reverse.com [168.1.18.1
30]
  4    22 ms    26 ms    21 ms  ae8.bbr02.eq01.syd02.networklayer.com [50.97.19.
152]
  5    21 ms    21 ms    21 ms  ae7.bbr01.eq01.syd02.networklayer.com [50.97.19.
60]
  6    68 ms    68 ms    69 ms  ae0.bbr02.ng01.per01.networklayer.com [50.97.19.
238]
  7    67 ms    67 ms    67 ms  ae7.bbr01.ng01.per01.networklayer.com [50.97.19.
232]
  8   115 ms   122 ms   116 ms  ae3.bbr01.eq01.sng02.networklayer.com [50.97.19.
235]
  9   116 ms   115 ms   116 ms  ae7.bbr02.eq01.sng02.networklayer.com [50.97.18.
171]
 10   181 ms   182 ms   182 ms  SFL-0022.asianetcom.net [202.147.33.13]
 11   183 ms   183 ms   183 ms  be2.wr2.sin0.asianetcom.net [61.14.157.185]
 12   252 ms   252 ms   252 ms  te0-1-0-5.gw2.lax3.asianetcom.net [61.14.157.210
]
 13   379 ms   381 ms   380 ms  xe2-3-1-0.gw1.lon1.asianetcom.net [202.147.58.15
1]
 14   343 ms   352 ms   344 ms  ccpgames.com [195.66.226.23]
 15   342 ms   342 ms   343 ms  srv202-c.ccp.cc [87.237.34.202]
Helios Anduath
The Scope
Gallente Federation
#196 - 2016-03-09 12:05:02 UTC
Panther2707 wrote:
Using a sydney server for VPN, first log is the route from me to the VPN, second log is from the VPN to the server. No issues at all last night, probably added another ~100ms to my ping but being australian im already used to high ping so its manageable until the issue clears.


Cool, that shows exactly what I would expect for a VPN that stabilises things - you take a completely different route in to the server.

Latency wise, it looks about the same. Your no-VPN pinplotter was showing 337ms, this tracert was showing 343ms so you have gained 6ms. That is not bad if it stabilises things.

I would suggest you keep an eye on things and try without the VPN once the cable faults in the area have been repaired.
Lichtenstein
Doomheim
#197 - 2016-03-09 21:37:21 UTC
Can someone recommend a VPN for me which I can use? I'm randomly getting socket closed messages and I want to test it out with a VPN.

Thanks in advance
pajedas
Doomheim
#198 - 2016-03-09 23:51:00 UTC
Lichtenstein wrote:
Can someone recommend a VPN for me which I can use? I'm randomly getting socket closed messages and I want to test it out with a VPN.

Thanks in advance

CyberGhost 5 seemed to work fine. The free version makes you wait... but it doesn't take long.

🐇
Shade Millith
Tactical Farmers.
Pandemic Horde
#199 - 2016-03-10 01:54:44 UTC  |  Edited by: Shade Millith
I've started watching the Resource Monitor.

As a client sits idle (Even with running modules) and is sending no data, the network activity it receives from the server will eventually drop down to 0. Even being removed from the list completely. Once it hits this point, it doesn't receive chats, convo invites, local changes, overview changes... anything at all from the server, until the client in question does a new action to send data and come out of idle. Coming out of idle can take upwards of 10-15 seconds.

This is not occurring to an actively used character.

I'm going to guess the socket closes are from not receiving any information from the server for too long, or a packet loss spike during that time of no data received. The problem with finding this out is I'm constantly getting small PL from certain servers along the route and always have, and the socket close notification isn't immediate.

I'll try a VPN in a bit to see if that changes anything.
Helios Anduath
The Scope
Gallente Federation
#200 - 2016-03-10 02:05:25 UTC
Shade Millith wrote:
I'm going to guess the socket closes are from not receiving any information from the server for too long, or a packet loss spike during that time of no data received. The problem with finding this out is I'm constantly getting small PL from certain servers along the route and always have, and the socket close notification isn't immediate.


Random packet loss at a hop or two along the route that does not propagate to the end is not necessarily an indicator of an issue as nodes that are loaded can drop ICMP responses to prioritise handling traffic. If you haven't seen it already, I posted some tips on analysing pinplotters here.

Can you post a screenshot of yours please?

Inactive clients dropping is definitely something odd and is probably worth a bug report eventually with as much details as possible.

Resource monitor is not the best way to keep an eye on network traffic for problems. If you really want to get technical, then Wireshark is your friend here as you can see what is being sent & received and whether there are retransmissions, errors, etc. It would need some setting up to only capture Eve traffic but it is doable.