[Solved by 3.3] Wireless devices can't see each other

Software SW help
1

I’m running into an odd issue where devices connected to wireless can’t see each other, but everything else works. That is:

  • Pinging a wireless client from another wireless client does not work (“no route to host”)
  • Pinging a wireless client from the router works
  • Pinging a wireless client from a wired client works
  • Pinging a wired client from a wireless client works

Oddly enough, if I reboot the router everything works for a while, which leads me to believe this is more likely a bug than a configuration problem.

Does anyone know what the problem might be, or how I can dig deeper to find out? Thanks!

2

Try to sniff the traffic on both ends.

Does it occur on 2.4 GHz, 5 GHz, both, or just between different cards?
Does the problem occur for both IPv4 and IPv6 or just one protocol?

3

Thanks, @Ondrej_Caletka!

I ran two series of tests, and got some odd results:

Laptop 5GHz, phone 5GHz: Fail
Laptop 5GHz, phone 2.4GHz: Fail
Laptop 2.4GHz, phone 2.4GHz: Success
Laptop 2.4GHz, phone 5GHz: Success

Laptop 5GHz, phone 5GHz: Fail
Laptop 5GHz, phone 2.4GHz: Fail
Laptop 2.4GHz, phone 5GHz: Fail
Laptop 2.4GHz, phone 2.4GHz: Success

It seems both devices on 5GHz consistently produces the bug, whereas both devices on 2.4GHz consistently works, with the remaining cases being flaky.

As for traces, I’m having trouble capturing any ICMP packets during the failure state. I’ll give it another try later today and report back.

4

Hey, just wanted to confirm the same problem. What’s even weirder, just reconnecting the stations makes this problem go away - I do this by having both 2.4 and 5Ghz networks the same name, security, etc. then disabling one of them - the clients will automagically roam. What’s more, this isn’t only on wifi - for example right now, i have a NUC on ethernet, and it can’t ping either my wireless printer or my Yamaha AVR, even though the omnia itself can. I think there’s something something bad going on on the bridge itself. Both the printer and the AVR are on 2.4Ghz, which seems to trigger the problem, but I’m not 100% sure, as I don’t have a station that is connected for long periods of time to 5Ghz

5

Hi, I also would like to confirm that I have similar problems, I have opened an other thread for it earlier I found this one:

6

I managed to get to a failure state with the laptop on 5GHz and the phone on 2.4GHz. This allowed me to test which interfaces the ICMP packets were going through, and I could definitely detect the requests (no responses) on both wlan0:

root@turris:~# tcpdump -i wlan0 -vv icmp
tcpdump: WARNING: wlan0: no IPv4 address assigned
tcpdump: listening on wlan0, link-type EN10MB (Ethernet), capture size 65535 bytes
19:39:51.314195 IP (tos 0x0, ttl 64, id 58277, offset 0, flags [none], proto ICMP (1), length 84)
    192.168.1.160 > 192.168.1.206: ICMP echo request, id 59154, seq 41, length 64
19:39:52.314166 IP (tos 0x0, ttl 64, id 27457, offset 0, flags [none], proto ICMP (1), length 84)
    192.168.1.160 > 192.168.1.206: ICMP echo request, id 59154, seq 42, length 64

and wlan1:

root@turris:~# tcpdump -i wlan1 -vv icmp
tcpdump: WARNING: wlan1: no IPv4 address assigned
tcpdump: listening on wlan1, link-type EN10MB (Ethernet), capture size 65535 bytes
19:40:05.341652 IP (tos 0x0, ttl 64, id 41383, offset 0, flags [none], proto ICMP (1), length 84)
    192.168.1.160 > 192.168.1.206: ICMP echo request, id 59154, seq 55, length 64
19:40:06.346173 IP (tos 0x0, ttl 64, id 54482, offset 0, flags [none], proto ICMP (1), length 84)
    192.168.1.160 > 192.168.1.206: ICMP echo request, id 59154, seq 56, length 64

Interestingly enough, after a while of keeping the ping command running, responses just started coming through. As soon as I switched the phone to the 5GHz network, though, it started failing again. And then after a while it started to work again, with occasional packet loss.

When things work, the ping times are high and irregular, going from < 100ms to over 1000ms, whereas pinging the router directly (192.168.1.1) is always on the order of a few ms.

So, to summarise all of my testing so far:

  • The problem affects both wireless bands
  • The problem is intermittent, and only starts happening after the router has been running for a while
  • It makes ping times high and irregular between wireless clients
  • It sometimes causes total packet loss between wireless clients, in that packets simply do not show up on the router (wlan0 or wlan1) and the ping command reports “no route to host”
  • It does not affect reachability or ping times between wireless clients and router
  • It does not affect reachability or ping times between wireless clients and wired clients
  • It does not affect reachability or ping times between wireless clients and WAN

All of this screams “bug” at me, but I’m at the end of my debugging skills to track things down. I suspect a driver or some firmware somewhere, but the kernel log is clean.

Could someone from CZ.NIC please comment on this issue? I’m happy to test as much as needed, but I need some guidance of where to look.

7

The irregular and high ping times might indicate it could be a power saving mode or issue either in the clients or in Omnia.

8

Hello again!

This may be a bit too early to tell, but it seems that update 3.3 seems to have done away with this problem. I’m cautiously optimistic for now, so I will mark the thread as solved, and change it back if the problem returns.