Help! CPEs acting strange

We have customers calling stating they can’t get online with our PMP320 gear.

We try to ping their radio and sometimes it takes a bit before they start responding.
When I’m finally able to log on it appears the radio should be running fine:
RSSI -71.89dBm, CINR 32.86dB, Uplink modulation qam64-ctc-5/6, Downlink modulation qam64-ctc-2/3

Service flows that I have setup on the AP are:
Default DL: HARQ disabled, 0 HARQ retries, Security disabled, DSCP Marking disabled, 25000000 Max sustained rate, 1 max traffic burst, 0 min reserved rate, 0 max latency, Data Delivery is Best Effort, Unsolicited grant interval is 0, Unsolicited Poll interval is 0.
Default UL: HARQ enabled, 2 HARQ retries, Security disabled, DSCP Marking disabled, 12000000 Max sustained rate, 1 max traffic burst, 0 min reserved rate, 0 max latency, Data Delivery is Best Effort, Unsolicited grant interval is 0, Unsolicited Poll interval is 0.

I have no idea what else to do. Do I have settings wrong?

What firmwares are running on CPEs and AP? Bridging?

We have one customer with similar symptoms, with the additional note that while RSSI & CINR look fine at the CPE end, the AP-end CINR for that customer fluctuates wildly and constantly, from 2 to 30 sometimes. Not yet found the cause of the fluctuation. (which has afflicted multiple CPEs at that customer site - about the only thing not yet ruled out is power inconstancy over 120ft buried cable) BTW - we’re using nrtPS/VR Data Delivery on uplink side of things, per Moto advice, but that shouldn’t be the cause of your problems I wouldn’t think.

Also, take a look at PerformanceMonitor->ActiveServiceFlows in the AP and see what degree of packet/byte drops it’s recording for that CPE, compared if possible to a known-good client on the same AP.

j

Running e2.1.2 on the APs and v7.1.3 on the CPEs with bridging enabled.

Looking on the AP under Performance Monitor > Radio Info the RSSI and CINR look fine. It’d be nice to see a max/min/latest type of stats for those though.

Looking at Performance Monitor > Active Service Flows it shows lots of the CPEs are dropping packets on the Uplink. The Downlink not so much at all really. Doesn’t this seem strange though since on the CPEs the uplink modulation is always perfect?

Min/Max/Last would be great, as would retention of data between reconnects, and tracking of session reconnects. Hopefully for the future, right? (I set up SNMP polling of rssi and cinr at CPE end polling every ten minutes, so we can see trends, but still - and there’s currently no access to that data at the AP end via SNMP)

Yes, it seems strange to have lots of drops with stable modulation, perhaps it’s too aggressive and should really be dropping down a level on the uplink modulation.

We’ve got this gear deployed at four towers now, same firmwares you’re running, and are less than thrilled with the products so far. Hoping (praying) that Moto pulls a rabbit out of the hat and the next firmware update addresses all this.

j

What OIDs are you using to track that data? I just started playing around with Zabbix so I’m going to try to add each of my 3.65 customers in to track their history.

Currently I’m polling a couple dozen OIDs every 10 mins, radiodbm and cinr are stuffed into an RRD as GAUGE values:

radiodbm, .1.3.6.1.4.1.10529.300.1.1.1
cinr, .1.3.6.1.4.1.10529.300.1.1.2

Then I graph them with dBm average as a line and CINR as area stacked on it. The result is a band graph where the height of the shaded portion at any moment shows the CINR at that time, the top of the shaded portion is the dBm and the bottom is the noise floor.

It’d be a huge improvement if we could poll per-cpe data from the AP, and if the AP retained data per-cpe between and across multiple connection sessions… :frowning: We have one customer where the uplink CINR keeps fluctuating drastically (2 to 30 range) and if I could poll that data from the AP then such anomalies would be readily apparent on the graph. (of course the SNMP MIBs for this gear are not exactly top-notch to start with: labels like ‘longtitude’ and ‘latititude’ for example…) hopefully Moto applies some efforts to this soon.

j

Max Latency should have a value of 200ms!

I do run Harq on my downstream but still experimenting with it.

Hi everyone,

i have the same Problem explained from pcpolo. someone has a solution?


Thx

We had a similar problem and found we were experiencing packet loss on the uplink after enabling HARQ and and working through retries we found that enabling HARQ with 6 reties resolved the issue.