That was both of us. And that's definitely the ethernet buffer issue on
the ApexPlus OMU. And I still can't fix it on HW ID 1 OMUs because
loading 3.1.5 will cause loss of in-band management (IBM doesn't respond
to ARP). 3.2.0 on the HW ID 2 works fine. And yes, the buffer issue was
much more noticeable on downstream 5ms framed radios. Enabling flow
control made the packet loss worse and did not improve throughput anyway.
I wonder if Jordan's problem could be a similar issue with the Mimosa
B5's? When I was debugging the Trango problem, I could go to the towers
and plug in with my laptop and pull 100Mbps no problem. Before we knew
about the buffer problem, we tried replacing Rocket PTPs with ePMP and
throughput improved. Then we got ePTP mode which is 2.5ms framing and
throughput got even better. Then the buffer issue was sorted out and I
was able to completely max out the ePTP links.
On 8/20/2016 4:32 PM, Ken Hohhof wrote:
I also seem to remember someone in my area like Nate or George posting
about problems with a Trango Apex Plus backhaul feeding an ePMP sector
and getting low throughput but not seeing the same problem with a 450
sector on the same backhaul. If I remember correctly they were kind
of pointing the finger at Trango, athough from the symptoms I wasn’t
sure that was clear since the 450 seemed to deal with it better. I’m
not sure if the OP had a router between the backhaul and the APs, or
just a switch.
*From:* Josh Luthman <mailto:[email protected]>
*Sent:* Saturday, August 20, 2016 4:26 PM
*To:* [email protected] <mailto:[email protected]>
*Subject:* Re: [AFMUG] New to list - issues (i think) with ePMP 2.4
I posted. Referred to Justin's issue from Wispamerica time frame.
Josh Luthman
Office: 937-552-2340
Direct: 937-552-2343
1100 Wayne St
Suite 1337
Troy, OH 45373
On Aug 20, 2016 5:23 PM, "Ken Hohhof" <[email protected]
<mailto:[email protected]>> wrote:
I think some of the comments about GigE interfaces might have been
misunderstood. I suspect whoever posted those was talking about a
situation where traffic gets bursted over one link and overflows
the buffers in the next device.
Also I would ask what hardware on the Mikrotik routers. I
remember having a problem with a Maxxwave ATOM box if I had a GigE
interface running at FastE (because the radio interface was only
10/100), it would exhibit severe packet loss. Downgrading to 5.x
firmware fixed that problem. I think that was also some kind of
interface buffer issue.
*From:* Chuck McCown <mailto:[email protected]>
*Sent:* Saturday, August 20, 2016 4:09 PM
*To:* [email protected] <mailto:[email protected]>
*Subject:* Re: [AFMUG] New to list - issues (i think) with ePMP 2.4
I think he has done that and it is AP to SM or AP to router that
is the problem.
*From:* Bill Prince <mailto:[email protected]>
*Sent:* Saturday, August 20, 2016 3:07 PM
*To:* [email protected] <mailto:[email protected]>
*Subject:* Re: [AFMUG] New to list - issues (i think) with ePMP 2.4
Then break it down. Put a mikrotik at the end point, and run tests
from there to all the intermediate points. Isolate the segment (or
segments) that are causing you grief.
bp
<part15sbs{at}gmail{dot}com>
On 8/20/2016 11:43 AM, Jordan Gregory wrote:
So here are the things i've tried.
Both my main Netonix Switch and the microtik switches has flow
control turned on. I disabled it and actually, if possible, the
speed got worse. So i re-enabled it.
I'm still unable to pass more the 3Mbps anywhere though.
I've been able to replicate this on every sector now. It has to
be something upstream from the access points, but I'm just not
sure where to look at this point.
Thank You,
Jordan Gregory
Founder / CEO
Hive Wireless, LLC
On Thu, Aug 18, 2016 at 10:50 PM, Jordan Gregory
<[email protected] <mailto:[email protected]>>
wrote:
Tik switch at the tower had flow control enabled on all
ports. I have disabled it now. Customer is asleep now i'm
sure, so I can't have them check. But I'll know more in the
morning. This was happening on all SMs though, so I may fire
one up just to see.
Thank You,
Jordan Gregory
Founder / CEO
Hive Wireless, LLC
On Thu, Aug 18, 2016 at 10:46 PM, Jordan Gregory
<[email protected]
<mailto:[email protected]>> wrote:
They aren't airfiber, they are Mimosa B5s
Microtik routers/switches.
I didn't enable flow control, but i'll double check
Thank You,
Jordan Gregory
Founder / CEO
Hive Wireless, LLC
On Thu, Aug 18, 2016 at 10:43 PM, George Skorup
<[email protected] <mailto:[email protected]>> wrote:
What is all of the equipment involved? Have you
looked for CRC or FCS errors on any ethernet links?
Unfortunately, the ePMP gear does not have ethernet
CRC errors stats. What about flow control on your
GigE interfaces?
Have you ran iperf from behind the SM all the way
through to your router?
You say 750Mbps. Are they AirFiber radios? Flow
control is what comes to my mind there. If you have
flow control enabled anywhere, turn it off and see if
end-user speed increases. We don't use it because
pause frames cause jitter for VoIP.
On 8/18/2016 10:34 PM, Jordan Gregory wrote:
Everyone,
I enabled the MIR profiles and set the one I am
talking about to 25000/5000, the link test shows
this speed perfectly, except when we do a speedtest
via speedof.me <http://speedof.me> or speedtest.net
<http://speedtest.net> or anything really, it's only
showing 3Mbps.
I have a cambium support ticket open, but that may
take a while.
I've dug through every setting I can find and can't
see anything that would be causing this.
Here is the generic layout of my network:
SM --><-- AP --> L3 Switch --> BH Radio --> <-- BH
Radio --> Switch --> Router --> Internet
I've tested the speed (iperf) all the way from my
router to my l3 swtich at the tower and I'm getting
roughtly 750Mbps avg to the l3 switch. I've tested
throughput on the switch as well and I was easily
passing 5Gbps through it.
I've seen no speed mismatch errors anywhere on the
network.
Thank You,
Jordan Gregory
Founder / CEO
Hive Wireless, LLC
On Thu, Aug 18, 2016 at 11:05 AM, Kerry
<[email protected] <mailto:[email protected]>>
wrote:
Here's a couple things to check,
Check the ap, is MIR under QOS enabled? if so,
there should be an entry under monitor->wireless
for the sm showing the rate the sm is 'queued' at.
The wireless link test will only show it passing
roughly what it is queued at. Also, check the
MCS column. If it;s a good link it will be at
the max Rate setting configured
under the radio config of the ap
On 8/18/2016 11:46 AM, Jordan Gregory wrote:
Good day list,
I have a Force 200 operating at very good
signal levels but for some reason, I can seem
to get any more than 3Mbps out of the connection.
Did i miss a setting somewhere?
Backstory: Brand new wisp
Thank You,
Jordan Gregory
Founder / CEO
Hive Wireless, LLC
