I don't think that that statement is correct, that a root with mac addresses: ##:##:##:##:##:A1 and ##:##:##:##:##:A2 the MBSS will always choose the A1 node as root? That wouldn't be very optimal given that A2 may be less hops to a particular node?
The way I understand it (and I may be wrong so correct me if I am) the MBSS doesn't determine a single global root to use. Mesh Roots continue to send out RANN announcements regardless of how many other root nodes are in the MBSS. And STA's determine the Root node to use based on metrics, but still retain proactive paths to each Root node. After re-reading the draft for mesh roots, and mesh gates. I re-gleaned the following, please let me know if im correct. 1) a STA can be defined as a Mesh Root and will then send RANN (root announcements) 2) a STA can be defined as a Mesh Gateway and will then send GANN (gate announcements) 3) the Root and Gateway STAs can be the same or separate nodes and there can be multiple of each if needed. 4.a) A non-root STA will create return paths to each Mesh Root after receiving a RANN (using PREQ and PREP handshake) Observation: so ideally (to reduce management traffic) unless some nodes are not able to communicate with a Root or the network is very large (geographically) so that nodes at each extreem have problems communicating to the other side you should only define 1 Root. when defining root nodes make them your best reachable STAs, i.e. high up on towers with excellent line of sight to majority of nodes. And roots should have good communications between each other if there are multiple. 4.b) A non-gateway STA will select a (default) gateway after receiving GANN by comparing metrics for known gateways. 4.c) Gateways will propagate their proxied (learned) external addresses with other Gateways and will forward externally destined communications to the Gateway that is currently proxying that specific MAC address. For example in 4.b above if the originating non-gateway STA sends its external destined communication to the wrong gateway (i.e. a different gateway is currently proxying the external address) the gateway will forward the external traffic to the correct gateway, the new gateway will then send a Proxy update to the originating non-gateway STA indicating to use itself for future communication with that external MAC address. If the external address is not known then the first Gateway will attempt to find the destination on the external network, failing that will send a Proxy error back to the originating STA. ??? at this point does the STA try alternate known gateways with worse metrics, or does communications just fail ? 5) When a path to a destination mesh STAs is not known the originating STA will select the best root path based on metric at that given time (based on existing proactive paths) to initiate communications, then will follow up with HWMP on-demand path mechanism to determine best route. David Fulgham On Tue, Nov 22, 2011 at 1:54 AM, Antonio Anselmi <[email protected]>wrote: > Ss far as I know the HWMP theory says that in case there is more than one > node configured as ROOT, a mesh node will receive proactive PREQs with > different originator mac addresses. > The mesh node then sets the lowest originator mac address as the ROOT. All > mesh nodes configured to act as ROOT, upon receiving proactive PREQ with > lower originator mac address, stop sending proactive PREQ and set the > originator of the proactive PREQ with lower MAC address as the ROOT. > So finally there remains a single ROOT node which sends proactive PREQs > and has paths to all nodes within the MBSS. > > Antonio > > 2011/11/22 MingAnn Ng <[email protected]> > >> Hi all, >> >> I had configured two mesh gate and 1 mesh point, they are connect >> together under 1 subnet. From Wireshark, two root announcement had been >> observed. >> >> How do we verify which mesh gate the mesh point used to ping the PC, >> which i connect to the two mesh gates using a switch. Does both of the mesh >> gate operates at the same time or only 1 of the mesh gate will operate at 1 >> time? >> >> Thanks, >> Ming Ann >> >> _______________________________________________ >> Devel mailing list >> [email protected] >> http://open80211s.com/mailman/listinfo/devel >> >> > > _______________________________________________ > Devel mailing list > [email protected] > http://open80211s.com/mailman/listinfo/devel > >
_______________________________________________ Devel mailing list [email protected] http://open80211s.com/mailman/listinfo/devel
