Gamers are complaining

RouterOS Wireless Networking
1

Hello All.

Here’s the quick list:
RB230’s loaded with level 5 2.9.12
Prism cards with latest firmwares (1.1.1/1.8.4)
3 cards active, fed to 15db 120deg sector antennas

All started off fine. I backhaul T1s to each of my sites, currently six. For months I advertised “faster than DSL” because… it was.

But now I have gamers that complain about frequent disconnects, poor ping times, and unexplained network drops.

Not much has changed with the configuration, except that we are using 2.9.12 and we have more users. Can anyone offer some experience or tips on how to optimize service for gamers. We don’t need much bandwidth (256 up/down with 1Mbit burst for 30secs) We use only B. And yes, I understand that game servers or the clients computers can be laggy for various reasons. i just want to make sure we are doing our best to optimize the MT routers for what we are offering.
Regards!

P2P Queueing causing VPN trouble....
2

How much suscribers do you have per card? Have you disable forwarding on each card? Post some config of your MT.

Regards!

3

Also, check if there is any p2p activity, if yes “drop” it…

4

802.11b could only handle max 3-5 mbps bandwidth (real time traffic) for point to point. If you use it on point to multi point, the max bandwidth would drop depends on the range from AP to client, the noise level, the signal level, many problems would affect wireless. :smiley:

Try to give us more information about your wireless, how many clients are on 1 sector, and some more details so we can help you. :wink:

Regards,
Lim

5

Ok, here’s some more info.

3 sectors, running 200mw cards
sector 1 has 13 clients
sector 2 has 9 clients
sector 3 has 44 clients

Now, before you say I have too many on sector 3, usage on this sector is very balanced (day/night) I rarely have more than 10 active connections going at once.

Average bandwidth consumption is 1.2Mb for entire sector

Just tonight, running pings from my router at home to my RB230 on the hill, 1/2mile away, I get anywhere from 4ms to 320ms

I am quite frustrated. Seems like the bigger we get, the worse our performance. I am biggining to suspect P2P traffic, but queues are set relatively low, and to the high bandwidth consumer, I make sure to keep their priority at 8, while "professional users get a 2 or 3. Plus, when I tried P2P/mangling, CPU usage shot to 100%, and still didn’t seem to affect things like bit torrent.

I link to the hill with a motorola 5700BH10, which has a near perfect connection (-41)

CPU usage averages 7%, sometimes peaks to 29%

I HAVE disabed forwarding on all cards.

All are on 2.9.12, as of last week. That is when I began noticing higher than usual complaints.

A year ago things were flawless. now, things suck. Please help. My rich competitor uses all Cisco, and LOVES to hear about our problems and has said some pretty shamful things about our choice in hardware/software.

6

here’s what my pings look like
Reply from 67.131.78.7: bytes=32 time=108ms TTL=63
Reply from 67.131.78.7: bytes=32 time=3ms TTL=63
Reply from 67.131.78.7: bytes=32 time=17ms TTL=63
Reply from 67.131.78.7: bytes=32 time=24ms TTL=63
Reply from 67.131.78.7: bytes=32 time=104ms TTL=63
Reply from 67.131.78.7: bytes=32 time=6ms TTL=63
Reply from 67.131.78.7: bytes=32 time=26ms TTL=63
Reply from 67.131.78.7: bytes=32 time=24ms TTL=63
Reply from 67.131.78.7: bytes=32 time=14ms TTL=63
Reply from 67.131.78.7: bytes=32 time=42ms TTL=63
Reply from 67.131.78.7: bytes=32 time=22ms TTL=63
Reply from 67.131.78.7: bytes=32 time=8ms TTL=63
Reply from 67.131.78.7: bytes=32 time=13ms TTL=63
Reply from 67.131.78.7: bytes=32 time=23ms TTL=63
Reply from 67.131.78.7: bytes=32 time=29ms TTL=63
Reply from 67.131.78.7: bytes=32 time=105ms TTL=63
Reply from 67.131.78.7: bytes=32 time=31ms TTL=63
Reply from 67.131.78.7: bytes=32 time=45ms TTL=63
Reply from 67.131.78.7: bytes=32 time=150ms TTL=63
Reply from 67.131.78.7: bytes=32 time=119ms TTL=63
Reply from 67.131.78.7: bytes=32 time=140ms TTL=63
Reply from 67.131.78.7: bytes=32 time=3ms TTL=63
Reply from 67.131.78.7: bytes=32 time=49ms TTL=63
Reply from 67.131.78.7: bytes=32 time=125ms TTL=63

Again, this is just a half mile away, adn the average ping time is 43ms.

I also found out something. Page resolution is VERY slow. I had DNS caching enabled before the 2.9.12 upgrade. Anyone have problems here?

7

Add this rule on terminal:

add chain=forward src-address=0.0.0.0/0 dst-address=0.0.0.0/0 p2p=all-p2p action=drop

and reboot, then check if the problem is still there, also check if connection tracking is enabled

Regards

8

is it possible to only block P2P from 7AM-11:59PM?
Also, I heard it was better to just let P2P barely leak through, instead of blocking it fully. What’s your take?

9

Yes you can, add this rule:

add chain=forward p2p=all-p2p time=7h-23h59m,sat,fri,thu,wed,tue,mon,sun action=drop

Regards

10

wow. so easy when you know what your are doing! I just tested this and it works great. Ok, one more variable. say I just want to give em 64k for P2P during the day, and unl at night. Is this possible?

11

Yes, add a simple queue like this:

queue simple add name="P2P" target-addresses=0.0.0.0/0 dst-address=0.0.0.0/0 interface=all parent=none direction=both priority=8 queue=default/default limit-at=0/0 max-limit=64000/64000 total-queue=default time=6h-23h59m,sun,mon,tue,wed,thu,fri,sat p2p=all-p2p

Regards

12

I’ll test this today. This community is lucky to have people like you.

As an added filter do you recommend limiting connections for subscribers? If so, how?

Thanks alot amigo!

13

You can try this one:

ip firewall filter add chain=forward src-address=0.0.0.0 dst-address=0.0.0.0 protocol=tcp p2p=all-p2p connection-limit=150,32 action=drop

Good luck :wink:

14

Hehe, well done cibernet
I couldn’t help im better :wink:

15

Thanks djape, BTW do you really want to catch 350 Mbits in the hotel room at MUM :smiley: :wink: just joking :stuck_out_tongue:

Regards

16

Hey Guys ! You are missing something…

I have made a 2.9.12 upgrade as well on “half” a radio chain (two out of four hops).

On 2.9.11 I had a ping of 4 - 5 ms to the gateway (with a load of 6 - 8 Mbps and all links in good condition on 5Ghz turbo). Now the pings average on 11 - 15 ms, which is the same we had before - deep into the internet.

The upgraded links seems to be a bit more stable in signal, but the upgrade definately had a negative impact - especially for the gamers.

We are NOT going to upgrade any more routers before this experimental wireless package is under control again ! However, I have considered the possibility, that 2.9.11 and 2.9.12 is slightly incompatible ???

Best regards
Kim C
Djurs, DK

17

I followed my idea about incompatibility between 2.9.11 and 12. It was completely wrong !

After upgrading the last two links to x.12, the chain became useless. Ping in the range 200 - 500 ms, dropped links, kernel faliures etc.

After disabling the wireless package and enabling the legacy package, followed by a reboot on all units, things became normal again.

Please MT - lets have the x.13 upgrade NOW !!!

Kim C

18

well, implemented the recommended filters. Works like a charm. Thanks man!

But it didn’t resolve my wireless issues. Ping times are all over the map, and page resolution is very slow. However, throughput seems to be right where I rate limit.

Tonight I am going to try downgrading a sector and evaluating performance. I’ll let you all know. I am still leary of this 2.9.12 and the effects it has had on Prism cards w/senao CB3+ clients.

Cheers!

19

How close are the sectors physically located to each other? do you have channel separation?

20

i just want to know how to set management bandwith dynamicly so if my custumer have problem whne their trafic full they can ping with small reply.

could you tell me howt to set bandwith management in mikrotik altought i used simple queue or queue tree n what a different from both, thank you.



regards :smiley: [/img]