Ошибка: open socket # left in connection

vaarsn nginx-forum на forum.nginx.org
Чт Янв 5 15:28:51 UTC 2017 

Имеется Nginx 1.10.0 включающий следующую конфигурацю:
----------------------------------------------------------------------------------------------
user www-data;
worker_processes auto;
pid /run/nginx.pid;

events {
        worker_connections 1024;
        multi_accept on;
        use epoll;
}

http {
        sendfile on;
        tcp_nopush on;
        tcp_nodelay on;
        keepalive_timeout 25;
        types_hash_max_size 2048;
        server_tokens off;
        send_timeout 2;
        reset_timedout_connection on;
        include /etc/nginx/mime.types;
        default_type application/octet-stream;
        access_log off;
        error_log /var/log/nginx/error.log debug;
        log_not_found off;
        ssl_protocols TLSv1 TLSv1.1 TLSv1.2; # Dropping SSLv3, ref: POODLE
        ssl_prefer_server_ciphers on;
        gzip on;
        gzip_disable "msie6";
----------------------------------------------------------------------------------------------

При рестарте наблюдаю в логах следующую ошибку:

2017/01/05 15:01:35 [alert] 1186#1186: *13388 open socket #15 left in
connection 17
2017/01/05 15:01:35 [alert] 1186#1186: *13441 open socket #45 left in
connection 22
2017/01/05 15:01:35 [alert] 1186#1186: *13444 open socket #48 left in
connection 26
..........................

Включил режим отладки, но толком тоже ничего:

2017/01/05 15:18:17 [debug] 1566#1566: epoll del event: fd:6 op:2
ev:00000000
2017/01/05 15:18:17 [debug] 1566#1566: epoll del event: fd:8 op:2
ev:00000000
2017/01/05 15:18:43 [debug] 1565#1565: post event 0000561F259CD460
2017/01/05 15:18:43 [debug] 1565#1565: delete posted event 0000561F259CD460
2017/01/05 15:18:43 [debug] 1565#1565: accept on 0.0.0.0:80, ready: 1
2017/01/05 15:18:43 [debug] 1565#1565: posix_memalign: 0000561F25A59750:512
@16
2017/01/05 15:18:43 [debug] 1565#1565: *5055 accept: 40.77.167.4:11355
fd:36
2017/01/05 15:18:43 [debug] 1565#1565: *5055 event timer add: 36:
60000:1483629583790
2017/01/05 15:18:43 [debug] 1565#1565: *5055 reusable connection: 1
2017/01/05 15:18:43 [debug] 1565#1565: *5055 epoll add event: fd:36 op:1
ev:80002001
2017/01/05 15:18:43 [debug] 1565#1565: accept() not ready (11: Resource
temporarily unavailable)
2017/01/05 15:18:43 [debug] 1565#1565: *5055 post event 0000561F259CDD60
2017/01/05 15:18:43 [debug] 1565#1565: *5055 delete posted event
0000561F259CDD60
2017/01/05 15:18:43 [debug] 1565#1565: *5055 http wait request handler
2017/01/05 15:18:43 [debug] 1565#1565: *5055 malloc: 0000561F25A79AC0:1024
2017/01/05 15:18:43 [debug] 1565#1565: *5055 recv: fd:36 242 of 1024
2017/01/05 15:18:43 [debug] 1565#1565: *5055 reusable connection: 0
2017/01/05 15:18:43 [debug] 1565#1565: *5055 posix_memalign:
0000561F25A4A620:4096 @16
2017/01/05 15:18:43 [debug] 1565#1565: *5055 http process request line
2017/01/05 15:18:43 [debug] 1565#1565: *5055 http request line: "GET
/robots.txt HTTP/1.1"
2017/01/05 15:18:43 [debug] 1565#1565: *5055 http uri: "/robots.txt"
2017/01/05 15:18:43 [debug] 1565#1565: *5055 http args: ""
2017/01/05 15:18:43 [debug] 1565#1565: *5055 http exten: "txt"
2017/01/05 15:18:43 [debug] 1565#1565: *5055 http process request header
line
2017/01/05 15:18:43 [debug] 1565#1565: *5055 http header: "Cache-Control:
no-cache"
2017/01/05 15:18:43 [debug] 1565#1565: *5055 http header: "Connection:
Keep-Alive"
2017/01/05 15:18:43 [debug] 1565#1565: *5055 http header: "Pragma:
no-cache"
2017/01/05 15:18:43 [debug] 1565#1565: *5055 http header: "Accept: */*"
2017/01/05 15:18:43 [debug] 1565#1565: *5055 http header: "Accept-Encoding:
gzip, deflate"
2017/01/05 15:18:44 [debug] 1566#1566: epoll add event: fd:6 op:1
ev:00002001
2017/01/05 15:18:44 [debug] 1566#1566: epoll add event: fd:8 op:1
ev:00002001
2017/01/05 15:18:44 [debug] 1565#1565: epoll del event: fd:6 op:2
ev:00000000
2017/01/05 15:18:44 [debug] 1565#1565: epoll del event: fd:8 op:2
ev:00000000


Вроде все работает, как и должно. Системные лимиты увеличил\снял:

root at thumborsrv:~# ulimit -n
64000

Добавил в /etc/sysctl.conf:
net.ipv4.tcp_tw_reuse=1
net.ipv4.tcp_tw_recycle=1

Параметры единственного виртуалхоста, который задействован на сервере:

upstream thumbor  {
    server 127.0.0.1:8000;
    server 127.0.0.1:8001;
    server 127.0.0.1:8002;
    server 127.0.0.1:8003;
keepalive 512;
}

server {
  listen 80;
  listen      443;
  server_name img.server.com;
keepalive_timeout 10;
add_header Strict-Transport-Security "max-age=63072000; includeSubdomains;
preload";
add_header X-Frame-Options "DENY";
ssl_session_cache builtin:1000 shared:SSL:10m;

 # block if specific token in header is not there for port 80
 # but allow requests from port 8080
 set $block_by_header 0;
 if ($server_port = '80') {
   set $block_by_header 1;
 }
 set $header_ok 0;
 if ($http_x_cf_secret = 'd41d8cd-98f00b2-04e980099-8ecf8427e') {
  set $header_ok 1;
 }
 set $block_key  $block_by_header$header_ok;
 if ($block_key = "10") {
   return 418;
 }

client_max_body_size 50M;
error_log /var/log/nginx/thumbor-error.log;
  # SSL certificates to use on secure connections
  ssl on;
  ssl_protocols TLSv1 TLSv1.1 TLSv1.2;
  ssl_ciphers EECDH+AESGCM:EDH+AESGCM:AES256+EECDH:AES256+EDH;
  ssl_certificate       /etc/nginx/ssl/img.server.com.crt;
  ssl_certificate_key   /etc/nginx/ssl/img.server.com.key;
  ssl_dhparam /etc/nginx/ssl/dhparam.pem;
ssl_prefer_server_ciphers on;
  # Forward requests to thumbor.
  location / {
    proxy_pass         http://thumbor;

  proxy_read_timeout     300;
  proxy_connect_timeout  300;

  # Default is HTTP/1, keepalive is only enabled in HTTP/1.1
  proxy_http_version 1.1;
  }
}

В логе самой тулзы вижу следующее:

[error] 1189#1189: *24 no live upstreams while connecting to upstream,
client: 41.182.178.150

И таких ошибок много. Помогите пожалуйста найти корень проблемы и устранить
его.

Posted at Nginx Forum: https://forum.nginx.org/read.php?21,271840,271840#msg-271840

Подробная информация о списке рассылки nginx-ru