i have clientes (AP mikrotik / Station various manufacturers) that have a very good signal strength (example: -65) and i ping them with big packets (ping xxx.xxx.xxx.xxx -t -l 1024) and the results are very poor… total packets lost are above 50%. those clients tell me that they have problems with connection.
those clients are connected from different distances… from 50meters to 1.5km
suggestions
I have seen similar responses and most every time it has been either a multipathing problem or is interference related. Would show me adequate signal strenght but ping replies were all over the place. A good connection should give you reasonably consistent replies to ping and no dropped packets. I wouldn’t expect ping replies to vary more than 5-10 ms any time, but that’s just from my own observations.
sounds like interference to me too, probably near your AP. Try changing frequency.
Also, what wireless card are you using in your mikrotik? There have been problems with R52H cards that stop transmitting at full power lately. Try using a Mikrotik box to connect to your AP, and check that both TX and RX signal strengths look good.
i have an rb532 with senao 200mw and a sr2
and a rb333 with two ubi xr2
it’s a tower with 4x90º sector antenas
Verify ping from 532 to client and 333 to client
Drop problem from board and panel sector
what frecuency use?
Drop problem on client, because one client with problem, down your system.
Greetings!
There are only two sets of four b/g mode frequencies in the US that are clear of each other:
1-4-7-10
2-5-8-11
…and somebody is bound to be using one of those (or one that is too close) in a populated area. My APs show log entries for clients disassociated with “extensive data loss” when one of my frequencies gets stepped on by another AP. This will cause slow response times on almost every protocol, not just ICMP, and loss of connection.
Just to clarify, “extensive data loss” could also mean the client powered off his CPE.
Powering down causes the extensive data loss log entry, but usually a one-time entry. A client on the fringe of your service area, or the frequency interference from another AP, will usually get several entries of auth-deauth cycles faster than I would expect from a power-down, tho.
As you use different client-CPE, you are not able to use Nstreme with polling. Sounds to me like a hidden station problem, speak some of your client-CPE’s can’t “hear” all of the others connected to the same AP, so they send their data simultanously, getting a “stop” from AP, waiting a short time and beginning to send again. Normally they listen if any other client is sending and wait till they have finished, but as they could not recognize the others (too far away, buildings between them etc.) they all send at the same time.
With standard 802.11b/g/a you will run into this sort of trouble getting worse with every client-CPE
Use Nstreme with polling and mikrotik-CPE for all clients to avoid this or change at least to 5 GHz to have more frequencies to switch between, and use smaller sectors (45° or 60°) to have the clients on one AP in a narrower location to each other.
Sincerly
Schnulch