We have created a network like enclosed picture.
Red circles are RB433 boards with two wireless cards, ROS 3.22.
White circles are RB433 boards with two wireless cards, ROS 2.9.51.
Red antennas are dual polarized antennas, 29 dBi.
What we call “Routers” in the picture are RB450 boards with 5 ethernet ports, ROS 3.22.
We created a full OSPF network, with MD5 authentication and network type=point-to-point.
Routers have redistribute default routes activated.
The “red” path cost is less than the other one, so data from server in site2 is sent generally through this path.
What happens is that OSPF has problems along red path (so by using all 3.22 devices), like strange errors in log:
Database description packet has different master status flag
new master flag=false
Discarding packet MD5 authentication failed
source = 10.0.12.2
Invalid cryptographic sequence number
mine = 39171
received = 39170
Discarding database description packet: wrong neighbor state
state = Down
Ignoring link state acknowledgment packet: wrong peer state
state = Down
Ignoring link state acknowledgment packet: wrong peer state
state = ExStart
and so on…
When the above happens (randomly), we cannot connect to site2, or ping are very high (2-3 seconds).
If we disable eth2 interface in Site2_ROUTER (so we disable the red path), Site2 is reached through the other link and works again.
We tried to use broadcast, NBMA, ptmp network types, with/without authentication with no success. Always same problems.
SNR is good for red path (about -70 dB).
I enclose also supout files of red devices.
Hope someone could give me any suggestions, as our customer is complaining a lot…
Thanks
supout.7z (725 KB)
