I recently set up (19 days ago or so) a new, non-exit, relay. It's gotten the guard, stable and fast flags, but it's consensus weight keeps jumping up and down for some reason (my understanding is that it should keep rising slowly after getting the guard flag). /var/log/tor/notices.log doesn't show anything of interest (I was getting an error due to too little CPU/RAM for the traffic load I was seeing, but I doubled the size of the VM after around 2 weeks, which got rid of that error).
Consensus weight is just an indication of the measured throughput of the relay. It's the "Observed Bandwidth", as measured by Bandwidth Authority relays. (For more on Bandwidth Authorities, see TorFlow.)
For example, if we look at your relay just now on Atlas, it shows:
And if we look at the descriptor of the relay, it shows:
r grumpytechie1 TKZOZ37P8jsTjP5wpqAulL3zqAU 1z/zwwjEv98DFWqRnF2vt4BnYL0 2018-02-27 18:03:15 184.108.40.206 9001 9030 s Fast Guard HSDir Running Stable V2Dir Valid w Bandwidth=7858 p reject 1-65535
We see that the
Bandwidth= value is the same as it's "Consensus Weight".
This is documented in
"w" SP "Bandwidth=" INT [SP "Measured=" INT] [SP "Unmeasured=1"] NL [At most once.] An estimate of the bandwidth of this relay, in an arbitrary unit (currently kilobytes per second). Used to weight router selection. See section 3.4.2 for details on how the value of Bandwidth is determined in a consensus.