Achieved it by symlinking the log to /dev/null
This alongside with nginx access log changes, and some other stuff related to cluster services and I have hourly writes on my root ssd's from approx 1.1gig per hour to about 50meg per hour.
Daily would be 26 gig down to 1.2 gig.
I havent touched rrdcached which I have seen mentioned on here as another source of heavy writes, although the access logs were been written to every 1-2 seconds, so much heavier.