It was mentioned, that packet loss is not expected when nstreme is used (or at least I understood it this way
However, I get consistent packet loss of 1-2% in a 14 hop link (example). The latency is not very high so I expect that none of the links overloads. Routers are ~40-50% loaded, which seems typical when this protocol is enables (700-800 MHz C3 with 2-4 Atheros interfaces).
585 packets transmitted, 568 packets received, 2% packet loss
round-trip min/avg/max = 20.131/79.849/551.363 ms
I can see packet loss when pinging the next hop router (at times).
What is causing packet loss with this protocol? Where to look at for improving packet loss?
Are you using revision 8 of Nstreme on all 14 routers?
If one of the links is unregistering and reregistering, then you can see a loss. If one packet was delayed by polling in versions ealier than revision 8, you could see it timeout if the timeout time was low (1s).
I am using revision 8 of nstreme everywhere in that part of the network (on around 30 MikroTik routers).
Is it correct to assume, that timeouts on nstreme revision 8 interfaces are only likely when the station disconnects from the AP. (all links are point to point) If the client disconnects from the AP there should be an entry at least in the APs log, but thre isn’t. Is it possible for both ends to dissassociate without log entry?
I am using the /ping command when testing packet loss to the next hop - what is the default timeout there?
