1) If you’re already patching nginx, why not just have nginx log straight into Redis? (Okay, this might be a bit difficult for non-programmers).
2) Nginx will happily and natively log to pipes (fifos). Might a simpler setup not be to log to /tmp/redis_fifo and run socat to take that fifo and direct it at Redis?
PS I’d never heard of empty_gif before, very neat, thank you.
And for anyone Googling for nginx syslog and see that there is native support, I’ll save you reading:
Logging to syslog is available as part of our commercial subscription only.