Community discussions

MikroTik App
 
i4ko
newbie
Topic Author
Posts: 39
Joined: Sat Sep 26, 2015 9:23 pm
Location: Arizona/Bulgaria/Sweden/France

Decrease in software quality from mikrotik?

Mon Jan 25, 2021 2:30 am

These are my observations on the last 2 long-term releases (6.46.7 and 6.46.8), there are things that just break that used to work before, and it is just annoying and frustrating.

1. False positive DFS detections
Observed on RB952Ui-5ac2nD that is located on the ground floor of internal courtyard (more like a chimney) of a building. The opening is approximately 15x18 meters, there are 6 stories on top of this router, and there are at least 8 walls (4 rebar, 4 wooden framing) in the nearest direction outside of this building, so there is absolutely no way that external radar signals can be seen at this install location. The closet weather radar is located some 17 miles away, and is actually below terrain level from this location. There are also at least 60 other APs visible from this one (only 2 are on DFS channels), and about 300 clients when I scan the site. Region is set to "united states" and tx-power is set to regulatory and mode is set to 5Ghz-AC only 20/40/80-XXXX, 802.11. U-NII-1 and U-NII-3 are so noisy that the AP does not select them. Usual selection after a restart is 5500 Ceee. Once or twice a day the radio will decide it has detected radar, again it is physically impossible for a radar to be received where this radio is.

2. Wrong/impossible channel selection
With the above setup, very often after the false positive the radar detection the radio will select a channel that does not exist in the united states AC channel plan https://apps.fcc.gov/kdb/GetAttachment. ... uROw%3D%3D , and should have never been selected with the above settings, such as 5710-eeeC, but also 5715, 5695, and other such that no client can see.

3. Not following scan list
If I set the scan list to 5270-5590 and 5650-5710, expecting that I would get channels 56,60,62,64,100,102,104,106,108,110,112,116,132,134,136,140, however that is not what the radio shows. It would say 5700-eeCe, which is channel 138 and clearly outside of the frequency I am asking the radio to work. If it selects 5700 it should be running in 20Mhz mode only as that is the only possible legal channel in US on that frequency with the above given limits

4. Bridges/bonded interfaces stop transmitting traffic when qos is disabled.
This is being observed on several RB750Gr3 with latest 6.46.8, the other side is hap ac lite or . There is a bonded interface (2 ethernet ports) part of a bridge, and also another ethernet interface. There are vlans under these that are part of separate bridges. There is a queue-tree with several queues, and more specifically 2 separate queues for the same traffic marker - one with parent global and one with parent the bonding interface. Ever since 6.46.8 as if the sub-queue for the market traffic that is under the bonding parent is disabled, the connection to the other router drops. Looking at the interfaces (vlans under the bonding) they do not show any transmitted packets, only received ones. That affects all vlan interfaces,not just those that the marked traffic relates to. With prior versions I could enable and disable either of those two queues at will without the things going lopsided.

5. NV2 not transmitting traffic
This used to work just fine, again not any more. I had several links at 2.4 with 5Mhz channels running NV2 with fixed downlink percentage of 20. These links are sending sensor data upstream, there isn't downstream. After client and AP upgrade to 6.46.8 (from 6.46.6 or earlier) all the links came up, but there was no data flowing from the remote sites. Look up the AP - see the link is up and running and station radios registered. Torch shows nothing coming in though, no RX packets on the interface. Several restarts I drive there with another AP, pour in the config - same thing. Drive out to the remote side and swap a client radio, I see station is connected but again no relived packets. Wasted 2 days in replacements until I decided to switch the NV2 to dynamic downlink - lo and behold - IP starts flowing. I switch back to the fixed downlink with 20% and IP stop flowing. Fiddle with percentages - set it to 26 - IP packets flow again.

I and my customers have invested a good amount of time in the mikrotik setups, and it is just not possible for me to recommend you any more when you break things that used to work just fine, especially when most of my customer's deployments are remote and take anywhere from 4 to 24 hours to get to. Couple of years ago this things used to happen in "Stable", that's when I switched everything to "Long-term", but come on, how can so many things, in so many different areas break, in what is supposed to be the most stable release? What is going on?

I have grudgingly started replacing some devices with other vendors.
 
mikruser
Long time Member
Long time Member
Posts: 578
Joined: Wed Jan 16, 2013 6:28 pm

Re: Decrease in software quality from mikrotik?

Mon Jan 25, 2021 1:36 pm

Yes, RouterOS has low quality.
I also have some issues ( viewtopic.php?f=2&t=171165 viewtopic.php?t=146665 )
MT support is also very bad - they refuse to admit there is a problem.
After replacing device to another vendor all problems disappeared.
 
User avatar
FToms
MikroTik Support
MikroTik Support
Posts: 87
Joined: Fri Jul 24, 2020 3:28 pm

Re: Decrease in software quality from mikrotik?

Wed Jan 27, 2021 11:27 am

1. False-positive radar detections can be triggered by sources other than radar or reflections of the AP's own signal. The last 2 LTS releases do not include any changes relating to DFS, so your observed increase in false positive radar detections is most likely caused by either an introduction of a new device/signal to the environment, a change in how the APs signal propagates through the environment or the AP choosing channels, signal conditions on which are more likely to trigger false-positive radar detections.

2. RouterOS supports selection of non-standard wireless channels, which may be useful in PtP setups. These channels are not 'impossible', though they may be 'wrong' for access points meant to provide network access for client devices which only scan standard channels. I will bring this issue up with my colleagues.

3. The scan-list parameter specifies frequencies to consider for the centre frequency of the control channel. In your given example 5700 falls within the 5650-5710 range. A scan list can be turned into a limitation to total occupied bandwidth by combining it with a set control/extension channel layout.

4;5 Going by what you've described in the post, my initial attempts at reproducing these issues were unsuccessful. Please report them (separately) by writing to support@mikrotik.com and including supout files.

Thank you for the detailed write-up and apologies for any inconvenience caused.
 
ste
Forum Guru
Forum Guru
Posts: 1924
Joined: Sun Feb 13, 2005 11:21 pm

Re: Decrease in software quality from mikrotik?

Wed Jan 27, 2021 12:03 pm

Some notes to the wireless part.

Radardetect/DFS is an increasing issue

- The used chipsets cant distinguish wireless signals from Radar in a lot of situations
- You could decrease the false detection rate using *correct* Gain-Data as this goes into the calculation
- With newer Firmware, Vendors have a higher number of false DFS-Events as they have to increase their detection rate of real Radar (older Firmwares are not doing this correct).
- With MT you could ignore regulations anyway as they still do no ATPC. So you send with much lower TX-Power (what happens using correct gain data) our you are illegal anyway.

We see this problem with all vendors. Some work toward a solution (prescan, second radio, ...) to avoid service interuption.

- nv2 is very old stuff which is no longer used.
- Where the hell ist 802.11ax ?
 
ivicask
Member
Member
Posts: 417
Joined: Tue Jul 07, 2015 2:40 pm
Location: Croatia, Zagreb

Re: Decrease in software quality from mikrotik?

Thu May 06, 2021 3:10 pm

Some notes to the wireless part.

Radardetect/DFS is an increasing issue

- The used chipsets cant distinguish wireless signals from Radar in a lot of situations
- You could decrease the false detection rate using *correct* Gain-Data as this goes into the calculation
- With newer Firmware, Vendors have a higher number of false DFS-Events as they have to increase their detection rate of real Radar (older Firmwares are not doing this correct).
- With MT you could ignore regulations anyway as they still do no ATPC. So you send with much lower TX-Power (what happens using correct gain data) our you are illegal anyway.

We see this problem with all vendors. Some work toward a solution (prescan, second radio, ...) to avoid service interuption.

- nv2 is very old stuff which is no longer used.
- Where the hell ist 802.11ax ?
For me its imposible to set mikrotik wireless link in legal way in most cases, i get up to 10 radar detects per day, if i set legal freq which skips radar detction than i get about 5 times less perfomance due colision with nearby APs..
Last edited by ivicask on Thu May 06, 2021 3:41 pm, edited 1 time in total.
 
pe1chl
Forum Guru
Forum Guru
Posts: 10186
Joined: Mon Jun 08, 2015 12:09 pm

Re: Decrease in software quality from mikrotik?

Thu May 06, 2021 3:23 pm

Yes it indeed is a common problem. It must be due to letters from the regulators to increase the DFS sensitivity. It affects different manufacturers.
Apparently the regulators and manufacturers don't understand that making the system unworkable will only result in users running ancient software or enable hidden workarounds to disable DFS.

Like the original poster, I recently had to do that. Installed a wAP AC indoors and put a scanlist with only channel 100,108 and 132 (staying clear of RADAR on 116 and 124 in this area) but it simply does not work, it sees RADAR on channel 100 (isn't there!) and hops to 108, sees RADAR there too and does not even consider 132.
Had to disable DFS to make it working.

Often the cause is that the AP is too close to users with a mobile phone. e.g. ceiling-mounted at 2.7m with users standing under it, or (as in case of the wAP AC) wall-mounted at 2m with users standing nearby.
(I often see that it only sees RADAR during business hours not during weekends, so it clearly is caused by users)
 
User avatar
mkx
Forum Guru
Forum Guru
Posts: 11383
Joined: Thu Mar 03, 2016 10:23 pm

Re: Decrease in software quality from mikrotik?

Fri May 07, 2021 12:20 am

Apparently the regulators and manufacturers don't understand that making the system unworkable will only result in users running ancient software or enable hidden workarounds to disable DFS.

Apparently regulators did not understand the reason for having certain frequencies reserved for special purposes and allowed incompetent manufacturers to pollute the spectrum with shitty RF transmitters. And after a few years of complaints from legitimate RF users regulators tried to fix the regulations only to find out that users already got used to use spectrum in a way it was never intended and that (some) users refuse to play by the book.
 
pe1chl
Forum Guru
Forum Guru
Posts: 10186
Joined: Mon Jun 08, 2015 12:09 pm

Re: Decrease in software quality from mikrotik?

Fri May 07, 2021 11:25 am

Apparently regulators did not understand the reason for having certain frequencies reserved for special purposes and allowed incompetent manufacturers to pollute the spectrum with shitty RF transmitters.
That surprises me too. How could anyone think that co-existence of an unlicensed transmitter system in the hands of consumers with a professional RADAR usage could ever work?
No matter what you try w.r.t. detection and avoidance, there will always be interference.
Of course such shared use of frequency space is common, but it hardly ever works out nicely. E.g. as radio amateurs we often see that parts of "our bands" are shared with others, and then we need to make sure that we don't interfere with them. But we are a small group of technical people, not a large group like WiFi users.
And the other way around, when unregulated users interfere with us (e.g. in the 433 MHz band) we are not allowed to disturb their operation.

Of course the problem is that spectrum is limited. There is not much that is free of any existing users and can be handed out to uses like WiFi, at least not in the usable part of the spectrum for this purpose (say, 1000 to 6000 MHz). A decision was made, and it turned out to be unfortunate.
 
User avatar
mkx
Forum Guru
Forum Guru
Posts: 11383
Joined: Thu Mar 03, 2016 10:23 pm

Re: Decrease in software quality from mikrotik?

Fri May 07, 2021 11:47 am

(I often see that it only sees RADAR during business hours not during weekends, so it clearly is caused by users)

I guess that's caused by Rx pre-amplifier not being able to lower gain enough ... which in turn saturates actual receiver causing all sorts of distortions. Those than can translate into spurious signals which trigger DFS.

While I was working for mobile network operator rolling out 3G (UMTS) network, we saw such phonomenon when the (measurement) phone was really close to antenna of base station (e.g. less than 5 metres). Even though UMTS features really good power control (while WiFi has none), phone still showed signal reception of a cell on same band but lower frequency (frequency offset was around 20MHz). It went away after we attenuated Tx of base station by 60 dB (it was a lab installation, hence that was the right thing to do anyway) ...

So yes, too little path loss between transmitter and receiver can cause all sorts of disturbances in receiver if dynamic range of receiver is not really large. WiFi TRXes being cheap sh..t usually don't (exceptions are probably only high-end professional APs).
 
pe1chl
Forum Guru
Forum Guru
Posts: 10186
Joined: Mon Jun 08, 2015 12:09 pm

Re: Decrease in software quality from mikrotik?

Fri May 07, 2021 1:10 pm

@mkx yes you are right, the receiver front-ends of WiFi APs are really sh*t.
Another issue is that we receive RADAR all over the band at an access point placed at 220m height in a radio transmitter tower, located about 20km from a weather radar.
It does not matter what channel is used, DFS detects radar everywhere. Likely a case of saturation of the receiver as well.
At that location I also notice that the APs receive other APs located at the same site (4 APs using sector antennas) at a fixed channel offset (I forgot what it was, I think 100MHz offset).

That kind of problem of course makes things worse than they need to be... software will have problems fixing those hardware issues.
 
User avatar
mkx
Forum Guru
Forum Guru
Posts: 11383
Joined: Thu Mar 03, 2016 10:23 pm

Re: Decrease in software quality from mikrotik?

Fri May 07, 2021 1:35 pm

Another issue is that we receive RADAR all over the band at an access point placed at 220m height in a radio transmitter tower, located about 20km from a weather radar.
It does not matter what channel is used, DFS detects radar everywhere. Likely a case of saturation of the receiver as well.
Weather radars are quite beasts actually. They transmit in short bursts and receive most of time (e.g. 1ms transmission, 999ms reception) and their antennae are really narrow-beam (norm is 1° in both directions). Their average Tx power is only a few watts, but then do the math: in those short time slots when transmitting their Tx power is a few kilowatts, "amplified" with antenna gain of around 45 dBi ... beaming into sh..tty wifi receiver.
 
User avatar
mkx
Forum Guru
Forum Guru
Posts: 11383
Joined: Thu Mar 03, 2016 10:23 pm

Re: Decrease in software quality from mikrotik?

Fri May 07, 2021 2:07 pm

As many of you could guess, the influence is not one-way (radars affecting wifi APs), stray wifi APs affect weather radar measurements as well. A weather radar image, showing the scale of the problem:

Image


Image shows measurements of otherwise "benign" atmosphere without any significant rain fall, so usually reflections would be blue at most. Blue mostly indicates clouds (without ground-reaching precipitation), green and up indicates actual rain drops (or snow fall) and red and purple colours indicate heavy hail.

The interfering AP is there for years and is identified to lay SE from radar (cross in the centre of the circle) way beyond nominal radar range (as noted on the picture radar range is nominally 250km and is drawn as change of gray background to darker shade).
 
pe1chl
Forum Guru
Forum Guru
Posts: 10186
Joined: Mon Jun 08, 2015 12:09 pm

Re: Decrease in software quality from mikrotik?

Fri May 07, 2021 2:19 pm

Yes, that is of course the reason for DFS!
The authorities and radar operators do not care if our WiFi network would be disturbed by radar pulses (and thus suggest us to go somewhere else)... what they care about is that we evacuate the frequency to avoid such disturbance of the operation of the radar.
But I think it is somewhat naive to assume that this would work. It requires only one or a few stations that are not compliant to have this interference.
 
User avatar
mkx
Forum Guru
Forum Guru
Posts: 11383
Joined: Thu Mar 03, 2016 10:23 pm

Re: Decrease in software quality from mikrotik?

Fri May 07, 2021 3:17 pm

As you wrote, the damage has already been done and the only thing remaining is damage control. Weather radars have been using their frequencies for decades and constraint is physics (reflection off water droplets) so it can't be changed (unlike air traffic radars). WiFi OTOH is technical problem and technically it would be failry easy to use 6GHz band instead of 5.5GHz. But as I said, regulators didn't think much when allowing co-existence between different users even though they have been warned before hand multiple times.

The history repeats itself with 5G and 14GHz band (passive! satellite measurements of water vapour). Etc.
 
Cablenut9
Long time Member
Long time Member
Posts: 542
Joined: Fri Jan 08, 2021 5:30 am

Re: Decrease in software quality from mikrotik?

Fri May 07, 2021 3:32 pm

I just make sure my equipment is the "International" version so I can select superchannel mode and never get any false positives in my APs. However, it shouldn't interfere with radar much because all the APs are in a location that is hostile to signals getting anywhere.
 
pe1chl
Forum Guru
Forum Guru
Posts: 10186
Joined: Mon Jun 08, 2015 12:09 pm

Re: Decrease in software quality from mikrotik?

Fri May 07, 2021 4:31 pm

WiFi OTOH is technical problem and technically it would be failry easy to use 6GHz band instead of 5.5GHz.
Unfortunately 6GHz is assigned to licensed fixed point-to-point networks here (in Europe).
While one can argue that this usage is outdated and mostly replaced or replaceable by fiber links, those users of course do not want to give up their spectrum either.
As the USA is already vacating part of this spectrum for WIFI, we can only hope that Europe will do a similar thing.
 
User avatar
mkx
Forum Guru
Forum Guru
Posts: 11383
Joined: Thu Mar 03, 2016 10:23 pm

Re: Decrease in software quality from mikrotik?

Fri May 07, 2021 6:20 pm

WiFi OTOH is technical problem and technically it would be failry easy to use 6GHz band instead of 5.5GHz.
Unfortunately 6GHz is assigned to licensed fixed point-to-point networks here (in Europe).
Exactly ... so when regulators are in doubt from whom to take, decision is easy: from the one who pays less.
It would be much easier to get co-existence of similar pieces of equipment (PtP systems and WiFi are both using some kind of digital modulation, at least pretend to have power control, can handle interference and transmission errors to some extent) than coexistence of much different types of equipment (such ad weather radars and wifi APs). But the crux of though process was: 90% of area is radar-free so only a few users would interfere with each other, OTOH PtP links are almost everywhere.

Who is online

Users browsing this forum: menyarito, phascogale, Soleous75 and 79 guests