I am in the process of putting up several new servers for testing. I am focusing on north east USA locations as they tend to have a good latency balance between USA and Europe. Also I believe Mach3 works well while CSS is not as stable? Sounds like we are more in need of a stable US server. If desired I can also put up a UK test server or two as servers near London tend to have the lowest pings for US players from Europe. This probably would drop US pings between about 15ms and 50ms (improvement will depend on player, I get 50ms less but average reduction would probably be closer to 20-25ms).
All of these are virtual machines. Some in the community believe Allegiance can't be stable on a virtual machine but I assure you this is not the case. That said, I fully expect some of these to not perform well which is the reason I have several servers to test. Latency should be good for all but need to make sure performance is stable under load.
Test Servers Online:
TestUS1 - Piscataway, New Jersey, USA [corgitech.com]
These servers are mostly untested so don't play anything important on them and they could spike during games. Keep me updated on how well they work. One or two server wide spikes on a server is enough to rule it out. Probably can eliminate some servers quickly and others will take extended testing.
TAG is not running on these servers during initial testing.
New Server Testing
Kudos for spending money on setting up another (test) gameserver.
If I may give a little advice. It might be wise to play on the test server with a small (say 3 vs 3) group of people while the server admin uses some monitoring tool to find out where the possible bottlenecks might be (network, hard disk, cache, program load etc etc).
If I may give a little advice. It might be wise to play on the test server with a small (say 3 vs 3) group of people while the server admin uses some monitoring tool to find out where the possible bottlenecks might be (network, hard disk, cache, program load etc etc).
At this point I am not looking to find bottlenecks. With the several servers (only a few if can't get running on Linux), I am hoping at least one of them will work perfect as is. Will worry about finding possible bottlenecks later if needed.peet wrote:QUOTE (peet @ Jan 28 2014, 03:29 AM) If I may give a little advice. It might be wise to play on the test server with a small (say 3 vs 3) group of people while the server admin uses some monitoring tool to find out where the possible bottlenecks might be (network, hard disk, cache, program load etc etc).
Definitely not going to upgrade TestUS2 anyway as it is more expensive than other options (and don't think host is worth a premium over the others). Most I can do with it is see about eliminating any unnecessary services.
Last edited by aem on Tue Jan 28, 2014 10:50 am, edited 1 time in total.
Yeah it seemed about the same for me. No reported packet loss and normal ping. But felt real unstable. Seems this server is pretty bad, but I didn't really have much faith in this particular one anyway. Probably no point in doing any more testing. Waiting on one Windows server to be provisioned and pkk to check if the IP was allowed on the other.Djole88 wrote:QUOTE (Djole88 @ Jan 28 2014, 04:44 AM) I just solo tested "TestUS2":
From Europe (Serbia) I have 135ms ping but the server spikes without reporting packet loss. I was trying to pu prox2 and it kept getting away
That is about the same ping I get on CSS but it doesn't spike there for me.
`yt AEM
Doesn't seem like I am going to get server working on Linux for now. Will have to try again later.