Linux – higher traffic priority for FPS game in Linux

iptableslinuxtctraffic-shaping

Goal: when running a FPS game in Linux, automate the setup of higher priority for its network traffic

I know such tasks are usually accomplished with a combination of iptables (to mark IP packets meeting certain criteria) and tc (to prioritize those IP packets).

Problems:

earlier version of iptables had an –pid-owner option with a warning in the manpage that this is broken on SMP kernels. My recent version of iptables (1.4.7) doesn't mention this option in the manpage at all
probably I can't use a single destination port to match traffic, because game servers run on different ports

I aim for a wrapper shell script which will:

run the game executable
find its name/pid
based on this will increase the priority of network traffic for this process
when I exit the game, will restore everything to defaults

Is this possible? I'm willing to go as far as dealing with custom netfilter modules, if there are any which could help.

Best Answer

I don't think you want to mess with QoS on the client; on your router maybe, but on the client machine, you're just introducing extra packet processing which is actually going to slow things down. On the router you're prioritizing those packets above other packets (like other http connections or other traffic also going through the router), so it's a win, but on the client you're doing extra filtering instead of just handing the packets straight through to the process.

You might want to use nice to adjust your client process priority though, so it will get faster access to the CPU than background things running on the same machine.

You might want to tune up your network receive/send buffer size to help minimize retransmits.

If you're trying to control actual bandwidth being generated by your computer, then you'll need more data on what exactly the FPS traffic is and how close it is to the 'other' traffic that you want to de-prioritize. You may be able to get away with something as simple as 'prioritize UDP over TCP' if your FPS uses UDP and your other traffic is all TCP. Or it may be more complex. Once you know how to characterize each set of traffic though, then you'll want to look at other answers for details.

Related Solutions

Linux – Linux Traffic Shaping Using TC

Assuming eth0 is a 100mbit Ethernet connection to the Cisco Router, it should be something like this (Isn’t it?):

tc qdisc add dev eth0 root handle 1: htb default 2
# 100 mbps
tc class add dev eth0 parent 1: classid 1:1 htb rate 100mbit
# To LAN traffic
tc class add dev eth0 parent 1:1 classid 1:2 htb rate 99000kbit ceil 100mbit
# IN traffic
tc class add dev eth0 parent 1:1 classid 1:3 htb rate 120kbit
# OUT traffic
tc class add dev eth0 parent 1:1 classid 1:4 htb rate 120kbit

# IN “bands” (one for each PC)
tc class add dev eth0 parent 1:3 classid 1:10 htb rate 24kbit ceil 120kbit
tc class add dev eth0 parent 1:3 classid 1:11 htb rate 24kbit ceil 120kbit
tc class add dev eth0 parent 1:3 classid 1:12 htb rate 24kbit ceil 120kbit
tc class add dev eth0 parent 1:3 classid 1:13 htb rate 24kbit ceil 120kbit
tc class add dev eth0 parent 1:3 classid 1:14 htb rate 24kbit ceil 120kbit

# OUT “bands” (one for each PC)
tc class add dev eth0 parent 1:4 classid 1:15 htb rate 24kbit ceil 120kbit
tc class add dev eth0 parent 1:4 classid 1:16 htb rate 24kbit ceil 120kbit
tc class add dev eth0 parent 1:4 classid 1:17 htb rate 24kbit ceil 120kbit
tc class add dev eth0 parent 1:4 classid 1:18 htb rate 24kbit ceil 120kbit
tc class add dev eth0 parent 1:4 classid 1:19 htb rate 24kbit ceil 120kbit

Would get me something like this:

+-----------------------------------------------------------+
|                      100 mbits (1:1)                      |
+---------+------------------------+------------------------+
| 99mbits |   120 kbits In (1:3)   |  120 kbits Out(1:4)    |
+  (1:2)  +----+----+----+----+----+----+----+----+----+----+
+---------+ PC1| PC2| PC3| PC4| PC5| PC1| PC2| PC3| PC4| PC5|
          |1:10|1:11|1:12|1:13|1:14|1:15|1:16|1:17|1:18|1:19|
          +----+----+----+----+----+----+----+----+----+----+

And for each band:

# PC1, IN
tc qdisc add dev eth0 parent 1:10 handle 20: prio
tc qdisc add dev eth0 parent 20:1 handle 22: sfq perturb 10
tc qdisc add dev eth0 parent 20:2 handle 23: sfq perturb 10
tc qdisc add dev eth0 parent 20:3 handle 24: sfq perturb 10

# PC1, OUT
tc qdisc add dev eth0 parent 1:15 handle 21: prio
tc qdisc add dev eth0 parent 21:1 handle 25: sfq perturb 10
tc qdisc add dev eth0 parent 21:2 handle 26: sfq perturb 10
tc qdisc add dev eth0 parent 21:3 handle 27: sfq perturb 10

+--------------------++--------------------+
|       PC1 IN       ||      PC1 OUT       |
+--------------------++--------------------+
|     PRIO (20:0)    ||     PRIO (21:0)    |
|      |      |      ||      |      |      |
| Prio | Prio | Prio || Prio | Prio | Prio |
|   1  |   2  |   3  ||   1  |   2  |   3  |
|(20:1)|(20:2)|(20:3)||(21:1)|(21:2)|(21:3)|
+------+------+------++------+------+------+
|  SFQ |  SFQ |  SFQ ||  SFQ |  SFQ |  SFQ |
|(22:0)|(23:0)|(24:0)||(25:0)|(26:0)|(27:0)|
+------+------+------++------+------+------+

And so on.

The rules should be like this

# PC1, OUT
tc filter add dev eth0 protocol ip parent 1:0 prio 1 handle 1 fw flowid 21:1
tc filter add dev eth0 protocol ip parent 1:0 prio 1 handle 2 fw flowid 21:2
tc filter add dev eth0 protocol ip parent 1:0 prio 1 handle 3 fw flowid 21:3

# PC1, IN
tc filter add dev eth0 protocol ip parent 1:0 prio 1 handle 16 fw flowid 20:1
tc filter add dev eth0 protocol ip parent 1:0 prio 1 handle 17 fw flowid 20:2
tc filter add dev eth0 protocol ip parent 1:0 prio 1 handle 18 fw flowid 20:3

and so on.

Any suggestion, comments, etc? (I have no experience in the field)

Linux – Minimal rate and default class problem for HTB

Since you didn't include the classifiers it's hard to deduct what traffic you exactly mean in each class. For instance, outgoing http or ssh traffic is very important for interactivity, incoming http not so much.

I would guarantee certain bandwidth for each service by saying: I have x kbps for incoming httpd traffic, and it gets divided equally among the users. If you have 10 or 100 users, it's fair." If you have high priority users or low priority users in each of these services you need to have additional classes and classifiers for them.

(Also I hope you know you can only shape outgoing traffic from an interface, NOT the incoming. That means if you want to limit the uplink, you have to work either with the outgoing interface to the internet or use the Intermediate Queuing Device. The lartc.org Guide is a very good resource.)

Best Answer

Related Solutions

Linux – Linux Traffic Shaping Using TC

Linux – Minimal rate and default class problem for HTB

Related Topic