I see here is 2 type of people on this forum, one whom this tweak is working, and one that isn`t.
I`ve seen this both cases and there are 2 cases for this error.
1. when you get this "upstream" error which is found in error_log, and this error you can easily solve by creating vhost_nginx.conf per domain or putting settings globally in nginx.conf
I`m thinking on this:
proxy_buffer_size 128k;
proxy_buffers 4 256k;
proxy_busy_buffers_size 256k;
but there is also another error which is happening after apache restart which brings us to case
2. When someone add subdomain, domain, or anything on hosting settings via Plesk, Plesk is doing apache restart.
Now in this case Nginx is having a connection to port 7080 which is Apache listening port and as soon Apache is down, boom, 502 bad gateway because there is no more Apache and nginx has no one to serve the request.
As soon apache is UP, F5, and all is up again.
I think that Plesk don`t have to restart apache for new configurations, reload is enough to apply all the settings because the way conf files are generated and included one in each other.
If they could just manage this problem it would be awsome, and I hope that now you will know why someone has problem solved and someone does`t but everyone using Plesk 11.X affected by this 2. problem.