[16985] in Athena Bugs
Re: strange inetd behavior on public workstation
daemon@ATHENA.MIT.EDU (Greg Hudson)
Sun Aug 1 14:47:55 1999
Message-Id: <199908011847.OAA20237@rcn-sucks.fsck.com>
To: Garry Zacheiss <zacheiss@MIT.EDU>
cc: bugs@MIT.EDU
In-reply-to: Your message of "Sun, 01 Aug 1999 10:35:38 EDT."
<199908011435.KAA04053@mary-kay-commandos.mit.edu>
Date: Sun, 01 Aug 1999 14:47:46 EDT
From: Greg Hudson <ghudson@MIT.EDU>
So, the inetd in question was in a state where accesson was 0 but the
switched services had not been switched off. I don't know how it got
into that state, but from there everything is explainable: access_off
had no effect because the inetd thought it was already in that state,
and when I ran access_on to reset the flag, inetd reinitialized the
switched services and lost track of the file descriptors with the
listener sockets (thus producing the "accept connections but hang"
state).
I couldn't figure out any way to get clues about how inetd would get
into such a state. The messages file had one relevant log: the
"write" service had been terminated for looping at some time in the
past.
