We have a couple routerboards with the latest OS and are using MT wireless cards as well. We set them up exactly as directed by someone to be used as a wireless backhaul. We have good signal and throughput. But every so often we’ll see the pings on these things start to incrementally go upward from 10ms all the way to 2000ms and then back to normal. This happens on and off during the day almost at regular intervals.
Then the router will stop responding alltogether, but continue to pass traffic. We need to reboot the routers manually to gain access again. Luckily we can do this remotely.
It also stops responding to SNMP requests. It is quite odd. This is happening across multiple of our backhauls we use. Some actually drop pings entirely, but when we ping devices behind the bridges they ping fine.
Anyone know if this is a common issue or does MT drop ping by default? These bridges have very basic routes etc, and basically route everything 0.0.0.0/0 inside our private network.
i have to ask, since you said snmp stops responding, are you seeing the memory graph climbing until it crashes ?
Sam
Nope, actually on the one bridge I can’t get snmp to work. Even though I’ve allowed 0.0.0.0/0 on it just like the others with read only.
Some of our AP’s are doing the exact same thing too. Some stop responding to ping entirely and we can’t get in with winbox or webgui, and it still passes traffic fine though. (until we manually reboot it)
I opened up two sessions in Winbox and watched as our pings went from 15ms to 50 to 100 200 400 800 etc.. eventually cutting out and then it would return to normal.
Cpu, mem on both sides were fine.
Noise floor was -90 and -93 and signal was good around -73.
(they both have Atheros AR5212 ABG in them)
But what I saw was the CCQ was dropping from 90% or so down to 30% during this. So it is a problem with some regular bursts of noise or something. We have I think 3ft parabolics up there and the link isn’t really that far. We’ve done much further. But ask the CCQ drops like crazy the noise floor remains the same.
Very curious.
Of course this doesn’t explain why we get locked out of the units alltogether eventually.
More information that is interesting - the pings and dropped sessions are only happening to windows machines. If I ping it from a linux router, or machine the pings are returned.
If I ping it from a DOS window in Windoze on any machine anywhere located around our network it will not return a reply.
This happens off and on over and over again throughout the day. Periods of 10 mins of loss of ping!
We are still having the same issues with even the latest firmware.
1km link 5.3 using two rb411’s 2 cm9’s -58 signal 36 SNR full connect at 54Mbit max noise floor I’ve seen is -94.
All of a sudden pings drop for 1 min at a time. Knocks people out completely. This is happening across many of our links. Tried different channels etc. Tried nstream and without nstream.. nstream makes it much worse!
Please help!!
PS - also once it reaches about 2000 pps or 15Mbit it chokes and dies..