source: http://zheng-ji.info/blog/2014/12/13/nginxcuo-wu-ma/

在定位线上服务问题的时候，通常会去查看Nginx 的 error log,
那么 error 的定义, 对查找问题就显得很有帮助

upstream prematurely closed connection
请求uri的时候出现的异常，是由于 upstream 还未返回应答给用户时用户断掉连接造成的，对系统没有影响，可以忽略.

recv() failed (104: Connection reset by peer)
服务器的并发连接数超过了其承载量，服务器会将其中一些连接Down掉；客户关掉了浏览器，而服务器还在给客户端发送数据;

(111: Connection refused) while connecting to upstream
用户在连接时，若遇到后端 upstream 挂掉或者不通，会收到该错误

(111: Connection refused) while reading response header from upstream
用户在连接成功后读取数据时，若遇到后端 upstream 挂掉或者不通，会收到该错误

(111: Connection refused) while sending request to upstream
Nginx 和 upstream 连接成功后发送数据时，若遇到后端 upstream 挂掉或者不通，会收到该错误

(110: Connection timed out) while connecting to upstream
nginx 连接后面的 upstream 时超时

(110: Connection timed out) while reading upstream
nginx 读取来自 upstream 的响应时超时

(110: Connection timed out) while reading response header from upstream
nginx 读取来自 u大带宽服务器pstream 的响应头时超时

(110: Connection timed out) while reading upstream
nginx读取来自 upstream 的响应时超时

(104: Connection reset by peer) while connecting to upstream
upstream发送了 RST，将连接重置

upstream sent invalid header while reading response header from upstream
upstream 发送的响应头无效

upstream sent no valid HTTP/1.0 header while reading response header from upstream
upstream 发送的响应头无效

client intended to send too large body
用于设置允许接受的客户端请求内容的最大值，默认值是1M，client 发送的 body 超过了设置值

reopening logs
用户发送kill -USR1命令

gracefully shutting down
用户发送kill -WINCH命令

no live upstreams while connecting to upstream
upstream 下的 server 全都挂了

SSL_do_handshake() failed
SSL握手失败

ngx_slab_alloc() failed: no memory in SSL session shared cache
ssl_session_cache大小不够等原因造成

could not add new SSL session to the session cache while SSL handshaking
ssl_session_cache 大小不够等原因造成