Recently, I updated nginx from 1.0.15 to 1.2.8, and find that the ports(shown by ss -s) increase much as below:
nginx/1.0.15
Total: 21696 (kernel 22773)
TCP: 111474 (estab 21422, closed 86149, orphaned 3803, synrecv 0, timewait 86145/0), ports 1417
nginx/1.2.8
Total: 21579 (kernel 22349)
TCP: 57466 (estab 21295, closed 32654, orphaned 3438, synrecv 0, timewait 32652/0), ports 11239
I updated nginx in order to use the ngx_http_upstream_keepalive module, most ports are used by nginx to connect with squid(port:8081):
(nginx/1.2.8) netstat -anp | awk '$5 ~ ":8081"' | grep -i time_wait | wc -l
10227
(nginx/1.0.15) netstat -anp | awk '$5 ~ ":8081"' | grep -i time_wait | wc -l
15
Is there any change that make nginx make lots of connections with squid and occupy so much ports?
nginx/1.0.15
Total: 21696 (kernel 22773)
TCP: 111474 (estab 21422, closed 86149, orphaned 3803, synrecv 0, timewait 86145/0), ports 1417
nginx/1.2.8
Total: 21579 (kernel 22349)
TCP: 57466 (estab 21295, closed 32654, orphaned 3438, synrecv 0, timewait 32652/0), ports 11239
I updated nginx in order to use the ngx_http_upstream_keepalive module, most ports are used by nginx to connect with squid(port:8081):
(nginx/1.2.8) netstat -anp | awk '$5 ~ ":8081"' | grep -i time_wait | wc -l
10227
(nginx/1.0.15) netstat -anp | awk '$5 ~ ":8081"' | grep -i time_wait | wc -l
15
Is there any change that make nginx make lots of connections with squid and occupy so much ports?