Hello,

we have used Nginx 1.4.x (the "extras", full featured package) for a year with no major troubles. Only some relatively rare error 403 we can't find the source of, it only happens when our XEN VPS (Ubuntu 12.04 LTS) is under (other sharing customers) load and our website is itself under high load.

I have upgraded to 1.6, the (Ondrej PPA). I had to downgrade to 1.4.7 VERY fast because error 403 (access forbidden) were literally spammed every other page just with a smidge of load.

The same content shows not a single error is used under light load so it's not a basic "wrong permissions" or similar issue.

I'd love if somebody could give me pointers about how to find out the real source of those errors 403. I can't replicate the issue on a basic "mule" computer, our setup uses a lot of features including varnish 3, GeoIP both at Nginx and PHP-fpm levels and so on.

The odd thing is, version 1.4.7 very seldom shows that spurious error, whereas to 1.6 it is a showstopper. To worsen things, sometimes varnish caches those 403 pages so unrelated users start seeing those errors as well.

I think the issue might be caused by too small buffers or something. The config file is massively huge so I don't know if it's a good idea to post it (nobody would care to read it all).

So I am just looking for pointers and ideas about how to pinpoint the problem source.

Best regards,
dfumagalli