RB532, CM9, and high pings

Hi everyone.

We were running an AP-2000 access point and seperate x86 MikroTik router at this one tower site, and it worked good.

We’re now trying out the RB532+CM9 MikroTik Router/AP combo box at that site, and, while I’m impressed at how much snappier my connection at home feels while connected to this AP/Router combo, I am seeing spikes in ping times to not only my wireless client antenna, but for other customers. (Ping tests were done from the wired end - which actually plugs into a 5GHz backhaul.)

I’m seeing ping spikes of up to atleast 125ms ever 3 - 5 seconds, somtimes it’ll go maybe 15 - 25 seconds before you see it spike like this.

I like to play Quake2, and I can watch the Quake2 “netgraph” to see when it happens. (It’s not the gameserver nor my route - its the AP/Router combo, because I also see these spikes when pinging the AP/Router combo from home - or from work on the wired side.)

My game just like everything else on my connection moves along alot quicker with this setup, which I love very much, however, these ping spikes cause my game (and everything else I imagine) to stutter for approx. 50ms (I’m guessing). It’s not fun for me nor my opponents to see that.

Otherwise, my connection to this tower is doing so much better… now if we can just iron out the ping spikes with it. :confused:

Ok, here are the details:

  • Wireless clients to this tower are all 802.11b
  • CM9 wireless card is running in ap-bridge mode set to 2.4GHz-B running on Frequency 2447 (Chan 8 I believe)
  • Supported B rates - I have all of them checked
  • Basic B rates - I have all of them checked
  • we’re not doing encryption, WDS, or Nstreme; we just have a 15 dBm omni antenna on the tower broadcasting this 802.11b signal.
  • we’ve got the wireless client antennas configured for static private IP’s, but the customer is configured for DHCP (gets a public IP)
  • we’re doing RADIUS authentication on the wireless and dhcp; we authenticate on the wireless by their antenna’s MAC address and we do the same for the customer’s PC/Router with DHCP.
  • the wireless interface (wlan1) has its ARP set to reply-only, and we’re putting the wireless client antenna MACs in as static ARP entries, while the DHCP server is setup to add the ARP entry.

Now, the ping spikes and/or overall connection quality seemed to improve some after I enabled all Basic B rates. (This was with all Supported B rates already enabled by default.)

Is there anything left that I can try out here to solve this?

I realized earlier that I didn’t have frequency-mode=manual-tx-power, when I had already configured tx-power-mode=card-rates and tx-power=18 (that of the 65mW CM9 card in 2.4GHz).. so I set that to manual-tx-power, and then attempted to lower on-fail-retry-time to something lower than 100ms, thinking maybe I didn’t have as good a link to the AP as I thought and that this was where most of my ~125ms ping times are coming from.

Unfortunately, it wouldn’t let me change to even 75ms, as well I’m still seeing these ping spikes here and there, sometimes with it acting like the connection’s bandwidth has been maxed out (basically starts acting like it can only recieve but not send very well.)

With our old setup, like I said, we used an AP-2000 access point from Proxim, usually with Orinoco PC Gold cards, which I think my boss said is a 100mW card… and that’s the conclusion I’m coming to, that maybe the CM9’s 65mW just isn’t enough juice for me.

That CM9 in the MikroTik AP is hooked up to a 15dBi omni antenna with an amp (though I don’t know how much power), and then I’m using a YDI/Terabeam/Proxim EtherAnt II Long Range antenna at my home (its an 18dBi directional panel antenna with a built-in amp, making the EIRP somewhere in the 40’s on this CPE, I think). The EtherAnt II gets atleast 10miles I believe.

So, do you think getting a higher power card like the SR2 would help us out?

After reading thru some of the recent posts here, I’ve tried the following:

  • Set my EtherAnt II-LR to ‘Automatic’ (only other options are 2 and 1Mb).
  • Disabling the QoS feature on my Linksys WRT54G router (home router).
  • Upgraded the RB532 Router/AP combo, 2.9.12 => 2.9.13
  • Disabled wireless package in favor of wireless-legacy (this resulted in no one associating to our CM9 - AR5213 card; I’ve since swapped it back to the wireless package and people are associating correctly again)
  • Changed preamble-mode from ‘both’ to ‘long’ (it says long is 802.11 compatible, but ‘short’ isn’t)
  • Changed from channel 8 to channel 11 (chan 11 is what our AP-2000 auto-selected for us when it was in use)

And after all of this, I think I still have this problem, though I will say it looks like its improved. (But who knows, it might’ve just improved because its so early in the morning – no one’s hardly using it, as I had hoped for, for reconfiguring this thing.)

This is my stopping point folks..

Any ideas ?

Err.. Let me clarify how the ping spikes are acting.

From a command prompt, pinging the MikroTik Router/AP you simply see anywheres from 60ms on up to atleast 120ms in every 1 of say 20 ping replies. The other 19 of 20 ping replies are all comfortably at 3 - 7ms.

It doesn’t look like such a big deal at first, but if you try to play a game like Quake with this, here’s what happens: you’re watching the in-game network graph - it spikes; well, ok, that one was minor; 15 seconds later it may do it again and this time you might see “hills” being drawn on the game’s netgraph and guess what - you’re barely able to move in the game until it clears up maybe a minute or two later. NOT cool to say the least.

So while I’m having a hard time even playing the game while that’s happening, I see the other players moving along just fine.

What’s going on here?

You could try locking down your clients datarate to lower rates. It could improve your ping.

This is the reason for your problem.

  • Basic B rates - I have all of them checked

You must set the basic-rate to only 1Mb/s – and don’t change anything that you don’t understand.

John

I turned on the extra Basic B rates, after the fact, in hopes of it improving. It has atleast made the connection seem a lot quicker. What can I say, I was desperate.

I’ll try turning it back down though.. Thanks.

Well, that didn’t do it either.

I managed to run a traceroute from a website while this was happening one time, and from those results, it appears to be a problem with either the backhauls or the MT router.

I had a friend traceroute me another time it happened, and he said it was just on the last hop – my Linksys router at the house.

Speaking of which.. maybe I’d better take my router out of the equation.

It’s not my linksys router. I’m wired up outside of my router and it still does it. This time when it happened, I went back to that website where I ran ping and traceroute tests to the MT router. Thing showed pings of 250ms up to 1150ms. (Pings were fine up until it pinged the MT router).

That tells me for sure that its either the backhauls or the MT router out there.

Currently, I can’t get access to the backhauls so I can’t check for that.
However, I’ve also got the router setup to graph its resources.. and I have noticed CPU usage on the router spike from about 12% to 48% sometimes while this was going on. (Didn’t happen the entire time the ping spikes were going on though.)

Heh, I was chasing my tail the whole time on this subject.

By the way, there’s only like 10 - 15 people on this access point.
Anyways, I’ll keep watching these two and update you.

It appears to be a problem with one of the backhauls - from the router I get a consistent 3 - 7ms to one of them, but the other one - when its acting up, soars to nearly 1000ms, and often hits 256ms on up to 500ms.

Geez, this was the last thing I would have considered. :blush: Especially since everything worked fine before putting in this router+AP combo box.

Though, come to think of it, I do remember seeing atleast one of the backhauls do this recently - just not on this scale. I’d imagine there’s water in a cable or something somewhere.

Sigh. Thanks for putting up with me.. :unamused:

For the record, my antenna’s signal to the AP is pretty darn consistent. I just wonder why my pings to the AP go up when the backhaul from that tower to the main one acts up, unless there’s more water than I thought …

There’s two parts to this:

  1. ping spikes (150 - 200ms) from me to the AP (from a normal 4 - 6ms); occurs atleast once every minute.
  2. difficulty sending but no problem recieving, accompanied by outrageous 1000ms pings; occurs atleast twice an hour and lasts for several minutes.

Last night, I had it figured to be the backhaul link that was causing #2. Maybe it is, maybe it isn’t.. I just know I’m not seeing it this morning. (While we were at it, we changed the frequency of the 5GHz backhauls.. maybe that’ll “fix” problem #2.)

As for #1, here’s what I’m seeing from work:

  • pings from my computer at work to the backhaul (A) up top: <= 1ms
  • pings from my computer at work to the backhaul (B) at the other tower: 1 - 9ms; usually 1ms
  • pings from RB532 at the other tower to its backhaul (B): 4 - 6ms (is this normal? I know older/slower machines have high pings from themselves..)
  • pings from RB532 to backhaul (A): 4 - 9ms

Then there’s this:

  • pings from my computer at work to my antenna at the house: 3 - 6ms; spikes as high as 150ms on average, atleast once every minute; sometimes more often/worse.

Problem #1 doesn’t appear to be the router nor the backhauls…just the wireless link from my antenna to the AP. But, my antenna’s signal (-55) seems pretty consistent.

So any thoughts? :frowning:

This is an old thread so ignore if your not reading.
Your wireless works on half duplex. One problem with your sporadic high ping time could be latency induced by the AP being busy sending or reiciving someone elses packets.
An isolated test would be to hide SSID, make sure no one else but you is Ack’d then try your ping test. I would suspect, all other things like noise and such being equal, you should find some pretty consistent times.

I’m having the sporatics high ping on 3 station(RB532+CM9) on the 2.4Ghz.
On the 5GHz I have no problem.

I have tried all, and at the end, I solved reducing the traffinc of my customers to 500 Kb max for access point.
Before of this setup I have used with success DLINK 2000AP+ with only the issue of the endurance but not for performance like as now I’m encountered.

I have same problem with CM9. Till he was dwl-2000+ everything was working good. I try to set mikrotik in client wds mode and d-link 2000+ in ap mode but no one cannot connect on d-link as long as he is connected to mokrotik via ethernet.

CM9 is working fine on ptp links but when it’s about AP for more clients it shows that it’s really bad card. with -56 % of signal ping goes over 300ms. I plan to buy SR2 and I hope that she will be better with AP on which about 10 clients are connecting, on 2.4

ktw-matt,

Have you solved your problem?

CM9 is working fine on ptp links but when it's about AP for more clients it shows that it's really bad card. with -56 % of signal ping goes over 300ms. I plan to buy SR2 and I hope that she will be better with AP on which about 10 clients are connecting, on 2.4

I have this think too, but I never use SR2 or SR52.
(MK team has not provided feedback)
I will wait your feddback before to buy other cards.
Meanwhile I will try other ways.

  1. Next week will arrive new 5006x chipset, I will try it as AP,
  2. also today I’m going to plug a mPCI card used on the 2000AP+ into RB523. I hope that the RB support Texas Instrument Chipset.
  3. I want to try INTEL chipset PROSET 2200 wireles mPCI of my laptop monted on the RB.

ropeba
are you Italian guy, right?

mail me public[chiocc]maxfava.com

Ciao

no, Bosnian :wink:

maxfava,

did you try plug a mPCI card from 2000AP+ into RB523?