Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!seismo!husc6!cmcl2!rutgers!uwvax!rhesus!uwmacc!uwmcsd1!lakesys!gryphon!cadovax!mitchell
From: mitchell@cadovax.UUCP (Mitchell Lerner)
Newsgroups: comp.unix.wizards
Subject: Re: Sockets stuck in FIN_WAIT_2
Message-ID: <1592@cadovax.UUCP>
Date: Fri, 12-Jun-87 18:14:52 EDT
Article-I.D.: cadovax.1592
Posted: Fri Jun 12 18:14:52 1987
Date-Received: Sat, 20-Jun-87 19:02:30 EDT
Organization: Contel Business Systems, Torrance, CA
Lines: 28


In article <1556@aw.sei.cmu.edu> pdb@sei.cmu.edu (Patrick Barron) writes:
>
>If I shutdown sendmail on my Ultrix VAX and there is still a socket out there
>bound to the local smtp service in FIN_WAIT_2 (waiting for a FIN from the
>remote host), then I'll never be able to restart mail unless I reboot the
>machine.  Does anyone have a program, or the proper "adb -k" incantations,
>to force a stuck FIN_WAIT_2 socket into CLOSED state?
>
>--Pat.

I seem to remember back when I worked on BSD type networking stuff that some
connections got stuck in a used or "zombie" state for long periods of time 
after a close() and and unlink() have already been issued.  

I dont know if this is directly related to what you are experiencing but the 
only thing that removed those zombie socket addresses from the address space 
(besides a reboot) was to do an "ifconfig Device_name -trailers down" (not 
sure about the args) and then ifconfig Device_name (i think).  Kinda a pain in 
the a** but it worked for me.

	- Mitchell


-- 
Mitchell Lerner
#  {ucbvax,ihnp4,decvax}!trwrb!cadovax!mitchell
#  cadovax!mitchell@ucla-locus.arpa