Too Many TIME_WAIT Connections

When a VPS or Dedicated hosting plan has sites that are performing poorly, an excess of TIME_WAIT connections may be contributing to the poor performance.

This article covers how to check and mitigate too many TIME_WAIT connections.

Related Article

Increase Apache Workers

Checking TIME_WAIT Connections

NOTE: While there are many traffic-related reasons why a website can have poor performance, at least one very easy test is to see if the poor performance is due to too many network connections stuck in TIME_WAIT.
TIP: If the server cannot immediately respond to a GET request due to sluggish MySQL performance, maxed-out IO, throttled CPU, or other resource usage-related issues, then the site visitor's connection status will change from ESTABLISHED to TIME_WAIT. An easy way to figure this out is to use the netstat command and count the number of TIME_WAIT connections.

1. SSH into the server as the root user
2. Runnetstat -nat | awk '{print $6}' | sort | uniq -c | sort -n
   EXAMPLE: In this example, there are way too many TIME_WAIT connections.  It is OK for TIME_WAIT to be 4 times the number of ESTABLISHED connections, but any significant amount higher than this will likely result in sluggish server performance.
   

         1 Foreign
         3 CLOSING
         8 FIN_WAIT2
        18 FIN_WAIT1
        24 CLOSE_WAIT
        25 LAST_ACK
        41 LISTEN
        84 SYN_RECV
       241 ESTABLISHED
      4668 TIME_WAIT

Resolving TIME_WAIT Connections

NOTE: If the output is similar to the above example follow these steps to mitigate this.

1. Log Into Root WHM
2. Navigate to the Apache Configuration section
3. Click Global Settings
4. Locate the Keep-Alive setting and select the ON radio button
5. Click Save
   TIP: After this feature is enabled, there should be an immediate reduction in TIME_WAIT connections, and within a couple of minutes the number of TIME_WAIT should be at an acceptable 1/4th ratio to ESTABLISHED connections.