Halo CE servers very laggy! Dont know why!

[Solaris 3,581]

From the video this seems more like a network/desync issue than a executable gone rogue.

 

It's possible this may have something to do with the way the NIC on the box is configured itself. IE, is the MTU set to low? Does the switch that the server is connected to have the wrong port speed associated with the port (ie, 100Mbps instead of 1Gbps or higher)? Is QoS incorrectly configured on the switch? Another server that's on the network could also be hogging bandwidth as well. It may be deeper than your server and you will have to coordinate with the datacenter that's hosting the box to find this out, as you won't have access to any of that hardware.

 

I would add the console parameter to the end of your dedicated server launch line and view the output of the dedicated server.

 

Edit: If you have antivirus running, make sure the server and directory have the appropriate exceptions.

 

Edit 2: After running Wireshark on the box, I definitely do see that there's some UDP packets being dropped (malformed packets). I was going to check the MTU size and enable Jumbo Frames, but then I realized that the NIC version was outdated (default shipped with 2008 R2). We're updating the NIC now to see if this fixes the packet drop issues.

 

Edit 3: They moved him from one switch to another to maybe see if it was congestion on the switch. It helped a little but the issue still persists.

Edited August 26, 2018 by Solaris

[Thomas 16]

This is odd, they did pings and lost nothing...  this is really frustrating...

[Thomas 16]

It turned out to be a peer issue.  A really big one at that.

 

I am not certain where or where the issue actually lies, but suffice to say that for the most part the issue has gone.  I do still get the odd lag spike, but it was nowhere near what it was when Solaris was assisting me.

 

I would like to thank everyone who suggested possible fixes in this thread, especially to Solaris who took the time out of his own day to help me figure out what the problem actually was, and after passing this data to

my hosts, was able to get them to try a few new things they never did before.  I saw a pretty dramatic improvement when they moved me to a new switch, and that seems to have cured the issue for the most part.

Edited September 3, 2018 by Thomas