All my clients on this AP are having this issue there are only about 10 subscribers with peak usage of around 30 mb with 75/25 gps sync. This test was ran during midday when usage is lowest. His MIR is 20/20 ..... Has anyone else had uplink problems ?
This looks odd. Perhaps there is interference which is affecting the uplink far more than the downlink. Do you see any dropped traffic in the uplink direction?
Hitting MCS-14 most of the time on uplink in that screenshot suggests it should reliably carry much more than 1.8mbps uplink... If it were interference I'd expect (hope) to see it dropping to a lower modulation, but almost 100% of the uplink traffic in your test went MCS13-MCS15.
Have you checked the AP MIR settings to make sure there wasn't a typo setting the provisioning at 20000/2000 instead of 20000/20000?
j
1 Like
If it is interfence it is not showing up on the scanner. There is no dropped traffic in the uplink direction .
It occasionally will hit 5meg but is most consistently 2meg, so i would assume the MIR is functioning properly. Here is a test from today and nothing has changed. It would appear to me that it is noise, but it was not dropping its modulation, or the radios were not reporting their modulation properly.
1 Like
Can you include a screen shot of the eAlign tool so we can see your CHAN1 CHAN2 signal levels? I've seen similar problems when one of the leads becomes damaged on the SM side.
One other thought, what is the firmware on both units?
Hi Ben,
When you say that there is 30 Mbps on the link, is this combined, DL, or UL. I am assuming you mean combined with maybe 23 Mbps DL and 7 Mbps UL. If the average SM is doing MCS 14 on the UL this is 90% of the maximum which is MCS 15 and if Link Quality (Uplink) was 78% which is what your first screen shot showed, then throughput would be reduced by 22% due to this percentage of packets failing. This would mean that of the maximum 25 Mbps UL available (if running at MCS 15) for 75/25, on average 7 Mbps is in use at MCS 14 (this is equivalent to 7.8 Mbps at MCS 15) and then having 22% failure means effectively 10 Mbps is in use of the 25 Mbps which is 40% of UL capacity.
I would think that an SM running at MCS 14 and Link Quality at 78% should achieve with the remaining 60% of the UL capacity during a link test of 15 Mbps * 90% (for MCS 14) * 78% link quality which is 10.5 Mbps.
If there is no other activity on the UL from other SMs, then I would expect 25 Mbps * 90% (for MCS 14) * 78% link quality which is 17.5 Mbps.
Therefore, I would expect your UL link test throughput to be between 10.5 - 17.5 Mbps. The second link test result you showed is what I would expect for a link test.
Therefore, your first result does not make sense to me.
Are you sure nothing else is going on in the UL at the time of the link tests?
Have you tried iperf using UDP from below the SM to above the AP? If so, what were these results?
We made a fix to the link test in 2.6.0, but it should not affect your link test results. Also, we made an improvement to the UL to protect from receiving unwanted packets (i.e. interference) if the interference is much lower in RSSI which should help receive management packets and facilitate general UL throughput. Perhaps you should upgrade to 2.6.0 and try this again.
If the UL MCS were a problem, it would shift down away from MCS 14 which it is not doing; so I do not think this is an issue.
I reran the link test on a local radio and received normal results in the UL.
Daniel Sullivan
ePMP Software Manager