How Do I Reduce the Time for Canceling TCP Connections in TIME_WAIT State on Linux

Question

In a large-scale Linux environment, a large number of Transmission Control Protocol (TCP) connections in TIME_WAIT state exist after Nginx reverse proxy service is adopted. By default, TCP connections in TIME_WAIT state are canceled in 2 minutes, which reduces system throughputs. How do I modify the kernel parameters of Linux to reduce the time for canceling TCP connections in TIME_WAIT states and increase the upper limit of nf_conntrack?

Answer

1. Log in to the Linux OS as the root user.

   Remotely log in to the server as the ossuser.

2. Modify the configuration file to modify the kernel parameters of Linux.
   1. Switch to the /etc directory.

      # cd /etc

      / is the root directory of Linux.

   2. You are advised to back up the configuration file sysctl.conf before modifying it. If the backup file name is sysctl.conf_bak, run the following command to back up the file:

      # cp sysctl.conf sysctl.conf_bak

   3. Open the sysctl.conf file in text mode.

      # vi sysctl.conf

   4. Add the following contents to the sysctl.conf file:

      net.ipv4.tcp_syncookies = 1 
      net.ipv4.tcp_tw_reuse = 1 
      net.ipv4.tcp_tw_recycle = 1 
      net.ipv4.tcp_timestamps = 1 
      net.ipv4.tcp_fin_timeout = 30 
      net.nf_conntrack_max = 655360 
      net.netfilter.nf_conntrack_tcp_timeout_established  = 1200     

      Set net.ipv4.tcp_tw_recycle to 0 in the NAT environmenHow Do I Check Whether Host Names in the SUSE Linux System Files Are the Samet and to 1 in other environments.

   5. Run the following command for the parameters to take effect:

      # /sbin/sysctl -p