Still learning about this myself but a couple of observations:
-
The age number in the Neighbor Table (the list of current 'single hop' neighbors being used) is evaluated differently than the one used in the Route Table Entry list (I haven't cracked that nut yet; I don't think I've ever seen any age other than 64 in that list). You probably won't see an age >6 in for an active link in the neighbor table; and it always will get reset back to 3 once link status reports are regularly received for an entry that has made it to age 5 (you will see this happen on your network if you refresh your browser page periodically). The only links that would show age <3 would be newly established ones that are considered in a 'probationary' period for which link costs are being evaluated. It's worth keeping in mind that the neighbor table entries aren't static and are always being re-evaluated (only the best 16 neighbors are used for routing for the first hop no matter how many routers are in your network).
-
incost and outcost and LQI may be calculated differently depending on what vendor's stack is running on the device in question. From what I have gathered they are computed from factors that include signal quality, link errors, and retries at the MAC level. Ultimately their purpose is to enable selecting the 'best' path to take (with lower numbers being better quality links, except for the special case of cost=0 which means 'cost unknown' for a newly establishing link, or 'stale link' for one which is ageing out). As such they aren't perfect metrics and the lack of standardization in their computation is part of the 'sport' in getting multiple vendors Zigbee equipment to play together nicely.
-
You aren't seeing all your devices because you aren't seeing the network's complete routing table, only the 'first hop' to the devices that the hub is currently targeting. The end devices that are reporting back to your hub would presumably appear in route table entries of the other routers throughout your network.
-
The device with poor LQI (that is reporting a higher LQI on the Zigbee logging page) is likely going through a repeater, and it is that 'last hop' repeater's LQI and RSSI that is being reported.
Not sure about the nulls, but I think they are artifacts of failed rejoin attempts. There's data exchange that occurs when the short address gets assigned during a join (or rejoin) and a null results when that process hasn't completed, afaik.
Edit 7-4-22: changed 'poorer' to 'better ' to correct categorization of Link cost
