* Simplify the OLSR watchdog
* Dont pull the routing table into LQM for supernodes.
LQM tracks routes on nodes to help keep leaf nodes connected even
when circumstances would probably prevent this. However on supernodes
the routing table is massive and pulling this into LQM will frequently
crash OLSRD. As we dont need this for supernode just dont do it for them.
- The "Always" checkbox now works for options with an explicit tag, and
is disabled for untagged options; the latter cannot be supported because
of a limitation in the OpwnWRT configuration language.
- Tagging by host name has been removed; it didn't work before and isn't
particularly useful in the AREDN context.
- Tagging by Agent Circuit ID, Agent Remote ID, and Subscriber-ID are
now supported so that a DHCP Relay Agent can be used to extend the LAN
across multiple subnets.
- Small improvements were made to error handling and hints.
* Reboot/reflash progress bar, time remaining in MM:SS
When rebooting or reflashing the node, display a progress bar as
well as the estimated time remaining. Also show the time remaining
in MM:SS format.
* Fix indentation error.
* Minor changes to improve browser compatibility.
Experimental wifi research reports that increased retries (rather than decreased as I originally
thought) improves tcp throughput on multi-hop wifi networks.
On some devices the hardware MAC for the WLAN and ETH are the
same so the 10.DTD and 10.WLAN addresses are identical. If we
detect this, bump the DTD address.
Commonly seen on TP-Link devices (but there are others)
Too many retries appears to cause network congestion issues slowing down
links with multiple hops, but reducing this to 1 then caused single links
to have poor performance. Try a rety of 2 as a better compromise.
* Revert hAP AC Lite xlink support
* Support CIDR (e.g. /24, /16, etc) on a xlink
Xlink had originally been envisaged as a point-to-point connection
leaving the "how" of that to other software. However, there's a use case
where the non-AREDN radios in the xlink need to be accessed, and allocating
addresses within the xlink's address range is a good way to do that.
By supporting a a subnet on a xlink we can enable this.
Tim Wilkinson