服务器网络重大故障—— LAST_ACK的socket有几万个

1. 服务器网络重大故障 : 处于 LAST_ACK 状态的socket太多了，且都是连接到同一个端口，导致服务无法使用。

最后经过几个小时的折腾，服务稳定下来了。

netstat -na |awk '{print $6}'| sort |uniq -c |sort -nr

4635 ESTABLISHED

3043 LAST_ACK

1001 TIME_WAIT

428 FIN_WAIT2

183 FIN_WAIT1

60 SYN_RECV

50 CONNECTED

23 CLOSING

15 LISTEN

8 STREAM

2. history|tac |head -100|sed 's/  /@/'|awk -F@ '{print $2}'

3.  history|tac |sed 's/  /@/'|awk -F@ '{print $2}'|sort|uniq -c|sort -n -k1 -r|grep -v kill

9 netstat -na |awk '{print $6}'| sort |uniq -c |sort -nr

6 redis-server /etc/redis/redis-18520.conf

4 vi /etc/redis/redis-18520.conf

3 netstat -ano|wc -l

3 cat /etc/sysctl.conf

2 telnet 172.17.0.1 6379

2 sysctl -w net.ipv4.tcp_retries2=5

2 redisc -h 172.17.0.2 info

2 netstat -an | awk '/^tcp/ {++s[$NF]} END {for(a in s) print a,s[a]}'

2 echo "net.ipv4.tcp_synack_retries = 1" >>/etc/sysctl.conf

1 watch -n 1 "df -i;df"

1 watch -d -n 1 netstat -ntlp

1 watch -d -n 1 'df; ls -Alt -F /root/.ssh/'

1 vi /etc/sysctl.conf

1 telnet 172.17.0.1 3306

1 /sbin/sysctl -p

1 nohup redis-server /etc/redis/redis_18520.conf &

1 nginx -t -c /etc/nginx/sites-available/default

1 ll /etc/ntp.conf

1 ll /etc/nginx/startssl_trust_chain.crt

1 echo "net.ipv4.tcp_tw_reuse = 1" >>/etc/sysctl.conf

1 echo "net.ipv4.tcp_tw_recycle = 1" >>/etc/sysctl.conf

1 echo "net.ipv4.tcp_max_syn_backlog = 4096" >>/etc/sysctl.conf

1 echo "25" >/proc/sys/net/ipv4/tcp_keepalive_intvl

1 echo "15" >/proc/sys/net/ipv4/tcp_keepalive_intvl

1 cat >> /etc/sysctl.conf