I'm still at it with the random no-hashing issue I keep having everyday with my Raspberry Pi & now 8 Gridseed G-blades, up from 6 G-blades, new 6FT USB Cables (down from 10 & 15 footers). And again, I even replaced the hub with a new Turcom 24 Port USB 2.0 Monster USB HUB W/ac Adapter Power Station.
I have only 2 g-blades running but i seen this often before, the only way i could get the 0 hashing away was after i turned power off on the failing one and power it back up.
minerd written by siklon (cpuminer-gc3355 v1.0e) has by far the best and most reliable built in automatic method of soft resetting stale chips, it's controlled via --gc3355-timeout switch, you just specify number of seconds after no share is submitted before restarting chips, in my experience works every time and it has yet to let me down.
The problem with this is that there is no way to know ahead of time what the pool difficulty is and how long it should take to find a share (unless a non-vardiff pool and you are doing the math). You could be resetting the device too soon, or waiting too long.
in the works for BFGMiner that does something similar, only it is based directly on the difficulty of the pool and works without additional arguments.