NeXusLAN Party Day 1 RRD Grapsh
-
Queue page for you as well.
-
You badass! but wtf is a RDD grapsh? :P
HFSC trick: On delay-insensitive queues use link-share [[b]m1=0,d=250,m2=<whatever "bandwidth"="" is=""></whatever>] on the queue and it will allow packets of that queue to be delayed up to 250ms (choose whatever worst-case delay you want). Since it is link-share it will only delay as long as it needs to.
You can use it to give downloads high bandwidth but huge delay, leaving low-delay opportunities for game/whatever packets. 500ms is probably a safe limit for TCP.
Like, m1=0,d=250,m2=50Mb would give the queue a 50Mb average bitrate with a worst-case per-packet delay of 250ms. Delay = time between HFSC receiving a packet's last bit (from system) and HFSC transmitting the packet's last bit.
-
Yes sorry . I had greasy pizza fingers while typing.
-
PRTG Graphs
-
Sunday Charts. Also some were complaining of some "packet loss" with League of Legends. Ran their trace program and here is snip of those results.
-
What all do you have PRTG setup to monitor sideout?
-
PRTG is monitoring PFSense , the core switch a Dell PowerConnect 2824 and my Vmware ESXi host. I have a laptop with PRTG on a static IP and output that to a large TV for attendees to see.
-
PFSense Queue view while LoL is going on so you can see traffic in the queue
-
Dropped ACKs. They're such small packets that they don't consume much bandwidth. Why not increase the queue size for those?
-
Mostly it was when I was playing around with a few things with M1 and D values on some queues. But it is something to consider.
-
Noticed some timeouts to Google DNS. Swapped in Level 3's . Here are the queues now.
-
Are you using DNS forwarding instead of Unbound? I rarely see timeouts to Google DNS, but I do sometimes see ping spikes, and not to both at the same time. So I many times have a ping -t running against 8.8.8.8 and 8.8.4.4, and I will see on start to lag, but not the other. Never for long and very rarely.
-
Yes I am using the DNS Forwarder in PFSense. Here is my DNS Benchmark.
-
Final Post for now then will post end of LAN stats later.
 -
Mostly it was when I was playing around with a few things with M1 and D values on some queues. But it is something to consider.
Any luck?
-
Good stuff, thanks for sharing.
-
I set the values and i noticed alot more dropped packets when I did that. I changed it back. Probably will need to do more testing at home.
Also think I was running into some issue with the NIC's. The server had Broadcomm NIC's and i put some customizations in the config file as per the PFSense docs wiki and I think I am going to go back to Intel NIC's .
I have 2 4 port Intel NIC Server cards. The old router had intel NIC's in it. It seems that PFSense likes the Intel NIC better .
Anyone have any issues with setting hardware offloading or ToE with NIC's?
-
If I remember correctly, ToEs do not honor traffic shaping because the ToE is the one sending the packets, not the shaper. I could be wrong.
-
Okay. I will have to test those settings as well. I saw the other post about Codel with UDP and dropping packets. Maybe that was some of my issue I was having.
Will have to test with putting UDP only queues under some other queueing and then using Codel for TCP only queues.
There were some complaints of packet loss in some of the games using UDP solely