1 reply to this topic

[SlowFuse]

First off, let me state that I love this game. Big Mechs and the BT world have been a fav of mine since the mid-late 80's. That said, the current issues with lag between the game servers and my client are about to make the game unplayable. I have read the forums many times over and see a good amount of other players experiencing similar issues, whether noting it in FPS or simply Ping rate.

I have waited to post anything until I could find a better overall data group to present regarding my situation. I have run traceroutes to any and all servers noted in my internet connection logs (via router) and cannot find one route that offers more than 33ms of hop latency. One server did have 11 hops, but even then ~300ms is not the end of the world (I'll just lead that Jenner a bit more!). That would lead me to believe that my standard "at rest" connection is fine.

Ping checks across any number of server locations across North America and even Europe deliver no more than 88ms latency. Again, a good sign given my overall connection.

The issue is that when playing the game, the Ping as noted in the player data menu (in-drop) will jump from its average rest of around 120ms (for me) to upward of 12000ms. This can occur while standing still with no movement from anything around me, or while moving and fighting. There is no one event that seems to trigger it. Today, I had all 4 of my homescreen mechs in matches at the same time, because each drop lagged out and presented the "Session Lost" message upon moving out of the drop zone. The client crashed to the login screen, and I was immediately able to relog, pick a new mech and launch a new mission while the last was still in progress.

Presumably, I could do this through the whole of my mech bay, but the first match would hopefully be done before I had all 16 mechs loaded in.

Each time this happens, I immediately go to my router status screen in IE browser window to see the link status and there is never an issue. NOTE: I keep the router status page open, so I can look quickly while testing the client relog. The connection log shows no state changes with regards to anything. No IP changes, no lost packets or irregular transmissions at all. The MWO client simply lagged out without any issues at the router / connection side.

I have reduced the system processes in Windows 7 to nothing other than what is needed. I have given MWO "real time" priority for processor actions as well as set all other non-required tasks to lower priority, although I highly doubt any of this is processor related. I am running an i7 with 8Gig ram and a ATI card with 2Gig vram. I did turn down all video settings to their lowest rates in game and hard-set FPS at 35 in the config file. All that works as intended, with my FPS never seeing any significant drop until the game lags out and I'm the only one moving of course. In actuality, I never see any issues other than this latency / lag one, and of course those dang Ravens chewing me up.

I live in an obscure area of the Colorado mountains where, admittedly, internet was not one of the best planned out infrastructure pieces built, but I have not once had the types of issues associated with MWO.

I play other online games daily, and am able to run 3 multiple clients simultaneously for some while also using VOIP (Teamspeak3). Worst case presented there is stuttering of voice on comms and this is a rare occurance. Unless I experience a total internet outtage or system crash, there is never any issue with dropping clients or lag that makes games unplayable like this.

It appears as though something in the chain between my location and the MWO servers is introducing high latency that is not showing during a standard trace / ping test while running a match, or that there is packet loss that is not being properly recovered from the server in the game code. It would appear that if the game doesn't get exactly what it needs across the net, it crashes, which is unacceptable. There should be some recovery routines that extend to higher levels of loss.

I see rubber-banding and that is an obvious recovery attempting to correct exactly the type of loss i'm mentioning, but is it selective? How high are the accepted losses before complete disconnect? What are the server loads like and how are connections prioritized? Do high-latency connections put a strain on the server and so get culled to prevent further issues across the match?

I don't expect answers to any / all of these, but something needs to be done. It seems that this game in particular is causing many people trouble with connections. There appears to be a least-common denominator here with regard to connection types, so I'm left to assume that those of us on the lower speed levels (mine's 1.5Mb max down in practical usage) are not accounted for in the code somehow.

Here's to hoping that you get this worked out with all the new "netcode" speak that going on, but it's more than just how the hitboxes work that is going on here. I would gladly accept a game that regesters the latency with my client and presents me with rubber-banding when necessary to regain my state, rather than the existing one that simply boots you and doesn't allow for any type of recovery to the match.


Thanks for the time!

Jae

[Phaesphoros]

Quote

Do high-latency connections put a strain on the server and so get culled to prevent further issues across the match?

I'm sure they've implemented something like that due to Host State Rewind, as they have to store the positions and states of all mechs up to a duration equal to the 2*the highest latency (or so).

I've also got some problems with server connection, but only when starting the patcher. It can take up to several minutes (no patch involved, just starting MWOPatcher.exe until the window shows up), sometimes it gets better after reconnecting to ISP.