Many CLOSE_WAIT connections

[cuihtlauac]

Recently, the server seems to become slower and slower until it becomes unresponsive. This only affects package documentation pages.

Connecting to the server and docker container shows there is an apparently always increasing number of TCP connections in CLOSE_WAIT status.

/ # netstat -tn | grep CLOSE_WAIT | wc
     3611     21666    292491
/ # netstat -tn | grep CLOSE_WAIT | wc
     3622     21732    293382
/ # netstat -tn | grep CLOSE_WAIT | wc
     3623     21738    293463
/ # netstat -tn | grep CLOSE_WAIT | wc
     3628     21768    293868
/ # netstat -tn | grep CLOSE_WAIT | wc
     3621     21726    293301
/ # netstat -tn | grep CLOSE_WAIT | wc
     3872     23232    313632

As a workaround, rebooting the machine will put the server in a working state. But daily reboots are needed.