Hello - apparently Turris Omnia here (with newest currently update) often breaks down.
I mean with this - only WLAN is breaking down. I then have to reset the device.
Then WLAN is working again. I don’t know why.
No offense.
Any details on the case would not hurt - it does not work - so consult - is not enough, sorry.
Potential duplicate of 5 GHZ network gone
I can also confirm that Wifi is flaky on Omnia, with 5.1.2 (was rock stable on 3.11) - could very well be the firmware IMO, I’ve seen that at least on stock OpenWRT new qca9880 firmware updates are rather common as of late (there’s a couple packages I like to have bleeding edge, mainly haproxy, so I do regular builds, and every once in a while cleanup my dl/ folder manually, that’s how I noticed).
Resetting the router generally works for me, but the next time my Wifi fail, I’ll try to grab some specifics.
To be fair, I’ve had to replace the Zenner diode and the electrolytic cap (after managing to plug my 27V speaker supply in the Omnia :D) at roughly the same time I did the 5.1 upgrade. But still, given that the thing works just fine even without those 2 components, I find it rather unlikely that it’s an electric phenomenon.
It seems to me, that WLAN of Omnia version 5.0.2 and 5.0.3 was more stable than
in current update of now with 5.1.x
So was WLAN in Omnia version 4.9.x more stable too than now.
I cannot reproduce this with ssh or similiar - because I have no password of turris for ssh.
Which wlan is breaking? 2,4 or 5 GHz? On mine, only 2,4 GHz (I am using WLE900 for this frequency). One of the problem is that it is kicking off only one phone several times per day. I solved this partly with setting channel from auto to specific one and the channel width to 20 MHz. But the problem occures sometime.
The second was that wifi has dissapeared or any of devices was unable to connect to it. In case that only restart helps, in system log, there are these messages:
[36866.247431] ath10k_pci 0000:01:00.0: wmi mgmt tx queue is full
[36866.253291] ath10k_pci 0000:01:00.0: failed to transmit packet, dropping: -28
[36866.260444] ath10k_pci 0000:01:00.0: failed to submit frame: -28
[36866.266473] ath10k_pci 0000:01:00.0: failed to transmit frame: -28
[36866.272813] ath10k_pci 0000:01:00.0: wmi mgmt tx queue is full
[36866.278661] ath10k_pci 0000:01:00.0: failed to transmit packet, dropping: -28
[36866.285820] ath10k_pci 0000:01:00.0: failed to submit frame: -28
[36866.291841] ath10k_pci 0000:01:00.0: failed to transmit frame: -28
I have contacted turris support and they answered, that there should be hw problem with the wifi card.
I am using CT drivers (can be found as alternative drivers in reForis or via LuCI) due to my second card, which is supported only by these drivers.
2,4 GHz was breaking down. At moment is peace again.
(5 GHz is not tested yet in Version 5.1.2 of Omnia.)
IP’s which were online - during these incidents - are (from Russia) :
fe80::c0b0:55ff:fe1f:44df%dummy0
fe80::2fe2:c42f:bdf9:8499%r_rmnet_data0
fe80::3a78:62ff:fe68:1844%wlan0
and BlackHat from Germany has :
fd80:43:b813:0:3a78:62ff:fe68:1844
fd80:43:b813:0:25db:6d8e:7609:7d81
(fdad:b10c:a::a8f:f4)
They seem to be eager to hack and crack turris devices …
It is you, who is joking …
Ummm, actually those are local address - just like 10.x.x.x and 192.168.x.x.
A local address is an address that multiple people can have at the same time - for example, almost everybody’s router has an address of 192.168.1.1.
The opposite isn’t true - for most IPs, there’s either exactly one piece of equipment with that address, or none.
So, if those are indeed hackers, they’re attacking you from your own home network - definitely not from somewhere else. maybe they have a Trojan installed on your laptop? 
You have overseen this : starting IP with fe80 is not local …
and in last line you can see IP with : fdad:b10c:a::a8f:f4 - this is an IP of a
so-called “Harting”-plug (query with google or duckduck tells so).
Here in my room is no “Harting”-plug.
And there is no Trojan too.
Well, I had not overseen it. fd**/8 are “unique local addresses”, and fe80 are even link-local.
To be clear, if you’re seeing them on WAN side of a router, they may not be that close geographically. And they might be malicious (say, other clients of your ISP). ULAs in particular could be on the other side of the world, at least theoretically, but none of them is publicly routable.