Killing CLOSE/CLOSE_WAIT Sockets?

[Dale Harrison <dth@ceg.net>]

Howdy,

Got a bit of a problem, not real sure how to solve it. 

Debian box used for mail. People pop their mail (via Outlook). Theres about 50 
clients that do this every 5 or so minutes. Every now and then people will 
complain that they can't get their mail [that in fact, they cannot be logged 
on to the mail server]

Theres nothing obviously wrong with the box [it's been happily chugging away
for 359 days] but I THINK the problem is this..

$ netstat |more
Active Internet connections (w/o servers)
Proto Recv-Q Send-Q Local Address           Foreign Address         State      
tcp        7      0 mail-i.proton.com:pop-3 122.122.125.125:1117    CLOSE       
tcp        7      0 mail-i.proton.com:pop-3 122.122.123.206:1059    CLOSE       
tcp        7      0 mail-i.proton.com:pop-3 122.122.123.20:1036     CLOSE       
tcp        7      0 mail-i.proton.com:pop-3 122.122.125.125:1045    CLOSE       
tcp        7      0 mail-i.proton.com:pop-3 122.122.123.199:1033    CLOSE       
tcp        7      0 mail-i.proton.com:pop-3 122.122.125.123:1047    CLOSE       
tcp        7      0 mail-i.proton.com:pop-3 122.122.125.125:1051    CLOSE       
...

etc. ~210 CLOSE or CLOSE_WAIT's. They don't seem to disappear. Is there any 
way I can nuke these manually? or decrease the time it takes for the kernel 
to clean them up? Or even increase the number of simultaneous connections per
socket?

Or am I barking up the wrong tree?

Help appreciated.. :)

D.
-- 
Dale Harrison  |                                          Universe, n.:
<dth@ceg.net>  |                                                  The problem.