Today I have face a big problem while using varnish with nginxcp + cpanel on a vps server .
Well the main site was working that was using a wordpress, however admin area because have a lot of plugins crashed with that stupid error.
So after a long search in ngnix log I find this error
upstream prematurely closed connection while reading response header from upstream
and this
an upstream response is buffered to a temporary file
However this seem’s that are not the problem for unavailable problem. However a fix for those error is to disable log in proxy.inc just add
proxy_buffering off;
However the fix is this in the varnish configuration default.vcl
backend default {
.host = “127.0.0.1”;
.port = “81”;
}
So after I modify this with
backend default {
.host = “127.0.0.1”;
.port = “81”;
.connect_timeout = 120s;
.first_byte_timeout = 120s;
.between_bytes_timeout = 120s;
}
The problem suddenly disappear . So basically the timeout should be increased for admin area if you have a lot of plugins there in wordpress and you are on a small vps.
