Ask the Experts: ePMP Troubleshooting

Our next Ask the Experts session will be Wednesday June 17 where members of the ePMP development and support team will share troubleshooting tips so that you can groom your network for peak performance. This session comes right after the Optimize Your ePMP Network webinars on June 16, so plan to attend the webinar and get more information on the discussion thread.

Bring your questions and comments, and be ready to learn new techniques to manage your network. 

ePMP Banner June 17.jpg

Participating is easy. Just login and join this thread at any or all of the following times on Wednesday, June 17th:

  • 8-9 AM CDT
  • 1-2 PM CDT
  • 9-10 PM CDT

Pass this information to your colleagues. If you are not yet a member of the Cambium Community, join at http://community.cambiumnetworks.com/ by using the link in the top right corner.

We look forward to the discussion and meeting with you. Add the sessions to your outlook calendar by clicking on the attachments below:

1 Like

Our ePMP experts will be joining us in 30 minutes. You can post any questions to this thread now.

Hi, i have raised the issue before "here" .

Today, we are effectivelly blind when looking for noise.

The spectrum analyser tool is a last resort measure, because of the very big downtime caused.

The percent-transmission time "per MCS" is nice, but is a very crude diagnostic when used alone, and today it is the only tool we have.

How do you suggest we procceed?

Good morning,

What would be the most efficient way to disable a client?

Currently with another vendor we go in and disable the LAN port, but I don't see that option on the EPMP.

Everything on the wireless side stays the same.

Thoughts?

Thanks,

Ryo


@Ryo Haines wrote:

Good morning,

What would be the most efficient way to disable a client?

Currently with another vendor we go in and disable the LAN port, but I don't see that option on the EPMP.

Everything on the wireless side stays the same.

Thoughts?

Thanks,

Ryo


Hi, With Release 2.4.3 (currently in Beta), you have the ability to disable the SM's LAN port. 

Thanks,

Sriram


@Guilherme wrote:

Hi, i have raised the issue before "here" .

Today, we are effectivelly blind when looking for noise.

The spectrum analyser tool is a last resort measure, because of the very big downtime caused.

The percent-transmission time "per MCS" is nice, but is a very crude diagnostic when used alone, and today it is the only tool we have.

How do you suggest we procceed?


Hi, 

I looked at your other thread and here are my comments. 

1. We provide retransmission counters (along with Capacity and Error drops) for every SM on the AP under Monitor->Performance. 

2. We have Quality and Capacity indicators which is a function of how well the link is performing. 

3. Tools->eDetect is a very useful tool to look for interferers around the AP and SMs. This will give you a good idea of the interference from other 802.11 devices on the channel. Running eDetect will not disrupt service. 

4. Lastly, the MCS stats is a good way to understand how your link is performing. Granted we don't have MCS spread per SM on the AP but we do provide current MCS per SM and that helps you determine how each link is performing

SNR is a good measure but it only reveals one piece of the puzzle. SNR is not CINR and hence it does not provide you information about interference, just the noise. So use it but we don't recommend that you rely on that measurement to determine the health of a link. 

Thanks,

Sriram

Regarding EVM, it is a function of overdriving the transmitter. On ePMP, we have taken care to not exceed Tx power on individual MCSs such that it would cause EVM issues. We will look into how an EVM stat can be provided but we know that the radio is calibrated and Tx power adjusted properly to eliminate EVM issues. 

i meant EVM in the interference evaluating sense, like in the ptp250 graphs / other vendor's constellation.

But thanks for the attention!


@Guilherme wrote:

i meant EVM in the interference evaluating sense, like in the ptp250 graphs / other vendor's constellation.


Understood. We will have to look into this. 


@Guilherme wrote:

i meant EVM in the interference evaluating sense, like in the ptp250 graphs / other vendor's constellation.

But thanks for the attention!


Question on this interesting topic - 

Would something like a trending graph showing per packet EVM (maybe represented by a simpler term) be beneficial ? This is very preliminary thought but something where you see EVM go up/down could help. Keep in mind that it may not track to our MCS changes as that would use a more sophisticated algorithm and this would essentially be raw packet EVM trend.

Sakid



Question on this interesting topic - 

Would something like a trending graph showing per packet EVM (maybe represented by a simpler term) be beneficial ? This is very preliminary thought but something where you see EVM go up/down could help. Keep in mind that it may not track to our MCS changes as that would use a more sophisticated algorithm and this would essentially be raw packet EVM trend.

Sakid


I think so, if that can be correlated to local noise to a higher degree than to good aiming of the antennas.

if it would be graphed in the eAlign window, we would see a stable RSSI line and a spike in the EVM line in case of spurious noise, right?



I think that showing the noise floor of every station in the "monitor > wireless" tab of the AP could be benefical too.

Not every station "sees" the same RF environment as the AP.


Changing subjects a bit, not every of our stations is perfectly aimed, especially because at the time of the initial deployment of the network, we had no eAlign tool, and the ATPC algorithm kept changing the MCS/TX power.  It would be very nice to have a "Alignment mode", where the MCS rate and the TX power are locked during the alignment procedure.




@Guilherme wrote:


Question on this interesting topic - 

Would something like a trending graph showing per packet EVM (maybe represented by a simpler term) be beneficial ? This is very preliminary thought but something where you see EVM go up/down could help. Keep in mind that it may not track to our MCS changes as that would use a more sophisticated algorithm and this would essentially be raw packet EVM trend.

Sakid


I think so, if that can be correlated to local noise to a higher degree than to good aiming of the antennas.

if it would be graphed in the eAlign window, we would see a stable RSSI line and a spike in the EVM line in case of spurious noise, right?



I think that showing the noise floor of every station in the "monitor > wireless" tab of the AP could be benefical too.

Not every station "sees" the same RF environment as the AP.


Changing subjects a bit, not every of our stations is perfectly aimed, especially because at the time of the initial deployment of the network, we had no eAlign tool, and the ATPC algorithm kept changing the MCS/TX power.  It would be very nice to have a "Alignment mode", where the MCS rate and the TX power are locked during the alignment procedure.




We will have to think through this a bit more. We actually need registration and packet transfers to capture that level of information but I see how localized interference representation at the time of alignment could help you. We do have eDetect that would actually give you same channel interference sources and this could be used in conjunction with alignment activities. As far as per SM interference level, we have link capacity & link quality as Sri mentioned before but maybe something like a DL interference metric that represents packet EVM could be useful. 

On your second topic, you can lock the max MCS rate and set the max tx power manually. Not sure how locking the MCS and Tx power to an exact level will help with the alignment. 

1 Like

Has anything been done or looked at into this issue?  http://community.cambiumnetworks.com/t5/ePMP-1000/75-25-ratio-low-client-s-upload/m-p/40407#U40407

Our ePMP development team will be joining us in a few minutes. Please post your questions now.

Also, if you like the response, be sure to click to give the post a kudo. If the response solves the problem, please click on ACCEPT AS SOLUTION.


@wvvamike wrote:

Has anything been done or looked at into this issue?  http://community.cambiumnetworks.com/t5/ePMP-1000/75-25-ratio-low-client-s-upload/m-p/40407#U40407


Yes, we are characterizing this behavior in our labs and hope to have an update either way for you.

Sakid

Hi.  With the 'eDetect' feature, I thought that was supposed to list all AP/CPE's that the unit could hear?  I understood it would include all widths, so if I'm in TDD Mode and 10 Mhz, I thought it would still show me the Clients' router (in 20 Mhz) if was visible to the SM?  I also thought that the eDetect would show all systems (Cambium, WiFi, UBNT) that it could hear.

The Site Survey tool is powerful, but having a eDetect that shows everything that the SM or AP can hear, and being able to sort it by frequency proximity and by RSSI, that's what I understood from the Cambium sales guys that it could do?

So, if it can't do that, then as a Troubleshooting help - what do we do when a client calls with slow internet, yet their signal looks fine.  How do we detect interferers which may be foreign or which may be on different channel widths?

Hi.  With our older (non Cambium) gear.  About 80% of our support calls for ''slow internet'' are really because the client has an unrestricted Torrent program that they don't understand, or they have devices with Cloud storage aps that they don't understand.  Right now, we can look at our AP's and see all stations and their throughput up/down in real time.

That allows us to see "at a glance" what everyone is doing on the AP, and every day, we probably have a dozen calls where people report 'slow' and we log in and they have 300 KBytes of transmit going on constantly that they are unaware is even happening.  That's out-competing their other traffic and they feel ''slow'' when it's really just 'occupied''.

So, with ePMP - how can I look at an Access Point and see a real-time. constantly updated list of all associated stations and their throughput?  I know we can't do it with our Canopy 900 gear (PMP100) and I don't see how to do that on the Cambium ePMP1000 either.

Hi ninedd and others. Our ePMP development team will be joining us in 30 minutes, so start posting your questions here. 

Dont forget to kudo posts that you  like, and also click on ACCEPT AS SOLUTION those posts that fix your problems.


@ninedd wrote:

Hi.  With the 'eDetect' feature, I thought that was supposed to list all AP/CPE's that the unit could hear?  I understood it would include all widths, so if I'm in TDD Mode and 10 Mhz, I thought it would still show me the Clients' router (in 20 Mhz) if was visible to the SM?  I also thought that the eDetect would show all systems (Cambium, WiFi, UBNT) that it could hear.

The Site Survey tool is powerful, but having a eDetect that shows everything that the SM or AP can hear, and being able to sort it by frequency proximity and by RSSI, that's what I understood from the Cambium sales guys that it could do?

So, if it can't do that, then as a Troubleshooting help - what do we do when a client calls with slow internet, yet their signal looks fine.  How do we detect interferers which may be foreign or which may be on different channel widths?


Hi,

eDetect will only give you what is running in the same channel and not in channels outside of your operating channel and bandwidth. Note that eDetect is running while your AP & SM is still in service. Unlike site survey tools it doesn't go into Rx mode only to scan all channels. When a customer calls in slow speeds and you are suspecting interference, you should run edetect and check what MAC adresses are the sources. We list the MACs of the interferers and their levels represented in colors. Its a pretty good start to the troubleshooting stage.

Sakid

eDetect uses free time within the frame and enters promiscuous mode.  This allows the SM to hear other 802.11a/b/g/n devices on the channel.  If the 802.11 interferer is not on channel, then it cannot be detected with eDetect.  5 and 10 MHz channels cannot be heard by a 20 MHz channel.  I believe 20 and 40 MHz channels can hear each other as long as the beacon information is on the same 20 MHz channel.

In order to hear other interferers (e.g. Canopy), the Automatic Channel Selection (ACS) tool could be used.  ACS can only be run in AP mode.  When ACS is enabled and run, it will scan for energy across the whole band and display the overall level on a normalized scale for each channel in the band and also pick the best overall channel for the selected bandwidth.  A person can use this information to for a variety of APs to perform channel planning or plan a single AP.