Hi guys,
I recently needed to upgrade my WiFi due to an upgrade in DSL speed, for which I chose the wAP ac RBwAPG-5HacT2HnD. I did not manage to unlock the full potential of the unit, as I have some weird link issues on the ethernet port. Here’s the physical network configuration:
Various WiFi clients
\\|//
wAP ac
|
FritzBox 7362SL (WiFi disabled)
/ | \\\
DSL Desktop PC (Realtek Gigabit) Other stuff on 100Mb
The Fritz has two Gb links and two 100Mb ones. As the wAP ac has Gb as well, I should a) be able to establish such a link to the Fritz, and b) exceed 8MB/s in data transfer from capable WiFi clients (say my laptop with the recent QCA9880 3x3 WiFi ac upgrade) to the desktop PC. Right?
Well, auto-negotiated links turn out to be either 100Mb or unstable Gb. Both units advertise 10/100/1000 speeds with Full and Half duplex settings. Currently, if I force 1000 FD or HD, I get nothing. There was a time when the connection would just randomly reset. When fully down on manual settings, cable test says:
Cable Pairs
shorted/21
shorted/22
open/19
open/18
(20m +3m cable for test, same with two 3m cables or direct plug into the fritz and 3m to the wAP)
“shorted” sometimes changes to “unknown”, rarely to “normal”, but “open” always stays open.
Attaching the very same cable except injector between laptop and Fritz does work just fine. netio 1.32tests (desktop is server, doesn’t really matter)
UDP connection established.
Packet size 1k bytes: 40.89 MByte/s (0%) Tx, 101.96 MByte/s (7%) Rx.
Packet size 2k bytes: 16.92 MByte/s (0%) Tx, 81.36 MByte/s (23%) Rx.
Packet size 4k bytes: 30.06 MByte/s (0%) Tx, 108.42 MByte/s (3%) Rx.
Packet size 8k bytes: 28.55 MByte/s (0%) Tx, 110.44 MByte/s (0%) Rx.
Packet size 16k bytes: 41.74 MByte/s (0%) Tx, 112.14 MByte/s (0%) Rx.
Packet size 32k bytes: 49.06 MByte/s (0%) Tx, 112.30 MByte/s (0%) Rx.
Done.
TCP connection established.
Packet size 1k bytes: 65.97 MByte/s Tx, 111.70 MByte/s Rx.
Packet size 2k bytes: 85.85 MByte/s Tx, 111.40 MByte/s Rx.
Packet size 4k bytes: 108.92 MByte/s Tx, 111.45 MByte/s Rx.
Packet size 8k bytes: 111.06 MByte/s Tx, 111.99 MByte/s Rx.
Packet size 16k bytes: 111.39 MByte/s Tx, 112.00 MByte/s Rx.
Packet size 32k bytes: 110.76 MByte/s Tx, 112.07 MByte/s Rx.
Done.
Same thing with the exact configuration that would plug into the wAP, except the PoE injector DC plug not being connected:
UDP connection established.
Packet size 1k bytes: 37.27 MByte/s (0%) Tx, 96.36 MByte/s (12%) Rx.
Packet size 2k bytes: 15.14 MByte/s (0%) Tx, 83.39 MByte/s (24%) Rx.
Packet size 4k bytes: 30.73 MByte/s (0%) Tx, 103.03 MByte/s (7%) Rx.
Packet size 8k bytes: 32.84 MByte/s (0%) Tx, 111.73 MByte/s (0%) Rx.
Packet size 16k bytes: 45.22 MByte/s (0%) Tx, 110.24 MByte/s (0%) Rx.
Packet size 32k bytes: 50.27 MByte/s (0%) Tx, 111.78 MByte/s (0%) Rx.
Done.
TCP connection established.
Packet size 1k bytes: 67.03 MByte/s Tx, 110.21 MByte/s Rx.
Packet size 2k bytes: 80.04 MByte/s Tx, 111.78 MByte/s Rx.
Packet size 4k bytes: 111.51 MByte/s Tx, 110.72 MByte/s Rx.
Packet size 8k bytes: 109.40 MByte/s Tx, 112.01 MByte/s Rx.
Packet size 16k bytes: 111.03 MByte/s Tx, 111.74 MByte/s Rx.
Packet size 32k bytes: 112.09 MByte/s Tx, 111.81 MByte/s Rx.
Done.
Cables are 1:1 Cat6 PiMF types from Reichelt, similar to this one.
So I’d say all cables are okay, the PoE injector is okay and does not harm performance, the Fritz and both of their gigabit ports are okay, also the desktop NIC is fine and the laptop (wired) NIC is fine.
Additional tests: I replaced the factory 24V/0.8A power supply with a generic one that now delivers 15V. No change (except whining noise of the power brick and a lower reported voltage in the health dialogue).
Ping test when in auto mode and there’s an unstable Gb connection:
Time (s) Interface Direction Src. Address Src. Port Dst. Address Dst. Port Protocol IP Protocol Size
13.196 ether1 tx 192.168.0.131 192.168.0.1 2048 (ip) 1 (icmp) 680
13.196 ether1 rx 192.168.0.1 192.168.0.131 2048 (ip) 1 (icmp) 680
14.201 ether1 tx 192.168.0.131 192.168.0.1 2048 (ip) 1 (icmp) 680
14.201 ether1 rx 192.168.0.1 192.168.0.131 2048 (ip) 1 (icmp) 680
15.2 ether1 tx 192.168.0.131 192.168.0.1 2048 (ip) 1 (icmp) 680
15.2 ether1 rx 192.168.0.1 192.168.0.131 2048 (ip) 1 (icmp) 680
16.203 ether1 tx 192.168.0.131 192.168.0.1 2048 (ip) 1 (icmp) 680<---
17.255 ether1 tx 192.168.0.131 192.168.0.1 2048 (ip) 1 (icmp) 680
17.256 ether1 rx 192.168.0.1 192.168.0.131 2048 (ip) 1 (icmp) 680
18.247 ether1 tx 192.168.0.131 192.168.0.1 2048 (ip) 1 (icmp) 680<---
19.355 ether1 tx 192.168.0.131 192.168.0.1 2048 (ip) 1 (icmp) 680
19.355 ether1 rx 192.168.0.1 192.168.0.131 2048 (ip) 1 (icmp) 680
192.168.0.1 is the Fritz, 192.168.0.131 is the ether1 address of the wAP, WiFi DHCP range of the wAP is 192.168.2.0 (side question: Is there a way to use DHCP relaying to the Fritz DHCP server?)
So there are quite a lot of cases where there is no RX for the ping. Long-term, there is about 1/4 to 1/3 packet loss, which is totally unacceptable. Same is true when pinging from any of the connected clients across the network.
Side note: Doing ping tests on the wAP brings up an error message when timeout is set to max (4999ms) and the first package is lost: “Couldn’t continue - object doesn’t exist (4)”. This does not happen on the default timeout or even higher ones like 2000ms.
When capturing the same stuff from with the Fritz, I get something like this:
Every TX gets an RX - however note that sometimes there’s an additional gap of half a second, while the wAP should ping somewhat precisely at a rate of 1 Hz. I’d need real-time timestamps to sync both up and see if the Fritz does miss some packets or the wAP dismisses some replies.
Next thing I’d like to test is connecting the wAP to the laptop via LAN and then route some traffic over WiFi through the wAP back into the laptop. If that fails, I’d say the wAP ethernet port is at fault. However, I’m not able to configure the unit that way, I have disabled the firewall rule so that I can access the unit over when connected to the Fritz, but I cannot get a direct laptop ↔ wAP connection to work. Fixed IP does work partially, the wAP can ping the laptop, but the laptop cannot ping back. I’m using the following config in /etc/network/interfaces on the machine:
iface mikrotik inet static
address 192.168.1.222
netmask 255.255.255.0
gateway 192.168.1.1
and start the network with ifup eth0=mikrotik. DHCP client settings in the wAP are disabled, and there’s an additional 192.168.1.1/32 address range on the 192.168.1.0 network, attached to ether1 interface. And that’s pretty much where I’m stuck at the moment.
Anyone got an idea of what to try next? Thanks! 