Dropping (nearly) all clients

[Dimos]

This is the second time that my server dropped all users with a german IP connection and only a swiss user didn't got dropped/lost connection.

Here is the log showing the client dropps. I also noticed while looking at the logs that the logtime is 2 hours of (should be 0:06 instead. Linux servertime is right).

2017-05-26 22:06:13.881287|INFO    |PktHandler    |1  |Dropping client 5 because of ping timeout 19 0 0
2017-05-26 22:06:13.881642|INFO    |VirtualServerBase|1  |client disconnected 'Censored'(id:16) reason 'reasonmsg=connection lost'
2017-05-26 22:06:13.882738|INFO    |PktHandler    |1  |Dropping client 6 because of ping timeout 19 0 0
2017-05-26 22:06:13.882777|INFO    |VirtualServerBase|1  |client disconnected 'Censored'(id:102) reason 'reasonmsg=connection lost'
2017-05-26 22:06:13.883105|INFO    |PktHandler    |1  |Dropping client 11 because of ping timeout 19 0 0
2017-05-26 22:06:13.883142|INFO    |VirtualServerBase|1  |client disconnected 'Censored'(id:452) reason 'reasonmsg=connection lost'
2017-05-26 22:06:13.891931|INFO    |PktHandler    |1  |Dropping client 14 because of ping timeout 19 0 0
2017-05-26 22:06:13.892006|INFO    |VirtualServerBase|1  |client disconnected 'Censored'(id:73) reason 'reasonmsg=connection lost'
2017-05-26 22:06:13.892329|INFO    |PktHandler    |1  |Dropping client 17 because of ping timeout 19 0 0
2017-05-26 22:06:13.892363|INFO    |VirtualServerBase|1  |client disconnected 'Censored'(id:80) reason 'reasonmsg=connection lost'
2017-05-26 22:06:13.892739|INFO    |PktHandler    |1  |Dropping client 19 because of ping timeout 19 0 0
2017-05-26 22:06:13.892780|INFO    |VirtualServerBase|1  |client disconnected 'Censored'(id:61) reason 'reasonmsg=connection lost'
2017-05-26 22:06:13.893105|INFO    |PktHandler    |1  |Dropping client 22 because of ping timeout 19 0 0
2017-05-26 22:06:13.893144|INFO    |VirtualServerBase|1  |client disconnected 'Censored'(id:72) reason 'reasonmsg=connection lost'
2017-05-26 22:06:13.893443|INFO    |PktHandler    |1  |Dropping client 23 because of ping timeout 19 0 0
2017-05-26 22:06:13.893541|INFO    |VirtualServerBase|1  |client disconnected 'Censored'(id:527) reason 'reasonmsg=connection lost'
2017-05-26 22:06:13.893792|INFO    |PktHandler    |1  |Dropping client 24 because of ping timeout 19 0 0
2017-05-26 22:06:13.893833|INFO    |VirtualServerBase|1  |client disconnected 'Censored'(id:361) reason 'reasonmsg=connection lost'
2017-05-26 22:06:13.894093|INFO    |PktHandler    |1  |Dropping client 25 because of ping timeout 19 0 0
2017-05-26 22:06:13.894129|INFO    |VirtualServerBase|1  |client disconnected 'Censored'(id:419) reason 'reasonmsg=connection lost'

The 'Censored' was added by me to hide the user names.

After that all clients with enabled autoreconnect reconnected between 2 and 50 seconds.

Can a DNS server/service cause this problem? Because this dropping started as I started using the ts.io service and all clients dropped using the ts.io dns to connect. The not dropped swiss user also uses the ts.io dns. Or is this just bad timing and not connected to each other?

Edit: while looking at the log again I noticed that 3 local hosted bots (on the same machine) also didn't got dropped/lost connection. So its looks more like an network error now.

[numma_cway]

ts.io also has a proxy service. That might get non-local clients dropped.
Using DNS (and not proxy) does not change anything compared to not using DNS, also because your server does not know anything about DNS. DNS is looked up once during connection and never used again while connected.

[Dimos]

ts.io also has a proxy service. That might get non-local clients dropped.

Yes, sorry forgot to give more infos. I'm NOT using the ts.io anti-DDOS (proxy) service. So I'm seeing my real server IP in the DNS resolve in the client log.

Using DNS (and not proxy) does not change anything compared to not using DNS, also because your server does not know anything about DNS. DNS is looked up once during connection and never used again while connected.

Thought the same, but wasn't completely sure about it, thanks for backing that up.

What I don't get is why only certain IP's (it seems like all german IP's for now) getting dropped. My server is located in Frankfurt(dedicated root server from netcup).

Is there any common linux log/command to see if there was another outtage on my server (not just the ts3server process)? Didn't found anything towards that in a quick google search.

[numma_cway]

That would be pinging/tracert'ing people all the time and checking for peaks/packetloss during outages. Note that if you too get dropped, SSH will close and cancel running programs. tracert/ping can be executed from home. For tracert'ing from client side, WinMTR can be used which will run indefinitely once started. It should be restarted every few minutes for better noticing peaks/packetloss. For ping, running ping -n 9000 1.2.3.4 will send 9000 pings for example.

There are differences in quality of datacenters, also in terms of routing. Your routing might just be bad. After I made that experience, I decided to spend a few more cents on decent routing and didn't have any problems for 4 years now.

[Dimos]

There are differences in quality of datacenters, also in terms of routing. Your routing might just be bad. After I made that experience, I decided to spend a few more cents on decent routing and didn't have any problems for 4 years now.

Well I think the same and already paying a fair price and hadn't had any issues in over a year of running the server on this machine.

Also the drops are random and the 2 I and the guild members noticed were in the last 9 days at completely random times, the first around 22:xx on a wednesday or so, not sure. And the second just now.

So permanently pinging/tracing will get funny