Hmm, yeah, I guess once Redis stops talking to the world, you’re going to have a hard time connecting to it to get stats… although if you run redis-cli while everything is hunky-dory, then stay connected, you should be able to keep running info stats (assuming Redis hasn’t completely lost its tiny mind).
Another thing to check is the Redis logs; from memory, I think we’ve got Redis logging to Docker, so ./launcher logs app is what you want there. If Redis is really playing up hard, that should tell you what’s what.
I’d also like to definitively rule out swapping as a contributing factor; can you collect the output of free -m from just before and just after the problem occurs, and run sar -W 1 (on the host) before and after 
.