I have about ten units in production, I have one that as soon as it gets exposed to the layer2 domain instantly reboots, replaced with a new unit & duplicate config, all good, 1 down, another unit is reporting no disk space available but has nothing in files and only the main package nothing else...2 down to bad hardware, Yes I tried the net-install non-sense that support immediately recommends for every problem... Fan bad? Net-install, Failed PSU? Net-install!

The other 8 units on 7.11.2 often will not respond to filter changes unless you reboot them, the other half of them will continue to announce bgp routes even when the peer has withdrawn or even killed the session (on either end, customer end or our end), We have fought these things as what I would consider Alpha testers for about 6 months now, I'm sure they will be a good product someday when the code is fully baked, I hear a lot of success lately from users but I really wonder what the use case is?

Unfortunately I think as a business decision I'm going to have to dump all of these things, it's unfortunate that CHR is the only alternative at high data rates but I don't have the hunger to 'fuss' with CHR/host for another 6 months to get everything optimized, I guess Juniper here we go! $$$