http/tcp server failing (looping) msg from inetd

[Scott Bernstein]

In the past few days, a problem has come up that I have never
seen before. My http daemon dies inexplicably, and then I get a message
looking something like:
Sep 15 21:38:10 hostname inetd[125]: http/tcp server failing (looping),
service terminated
in my messages file.

This is running on a sparc w/SunOS 4.1.3. I was running the latest NCSA
daemon,
but decided to try the Apache daemon after the problems started. Seems like
the Apache didn't make any difference since it's still happening.

If I do a kill -HUP on the inetd process, everything comes back normally (but
occasionally dies again). This was happening infrequently, but seems to
be picking up in frequency.

Any ideas on why my httpd is crashing? What does this message mean, and why is
my http server looping? Solutions? any clue at all?

Scott Bernstein
s...@questtech.com

[Kevin Kadow]

In article <43d1hf$c...@news.ios.com>, Scott Bernstein <sb> wrote:
>In the past few days, a problem has come up that I have never
>seen before. My http daemon dies inexplicably, and then I get a message
>looking something like:
>Sep 15 21:38:10 hostname inetd[125]: http/tcp server failing (looping),
>service terminated in my messages file.

This is a "feature" of the Inetd program- if it sees too many requests for
a service within X minutes, it disconnects the service for 10 minutes or so.

The timeout and wait time are compiled into inetd so there's no easy way to
change them.

The solution is to run the httpd as a standalone daemon instead of having
inetd start an instance of it each time a request comes in.
--
kad...@msg.net Kevin Kadow
Yes, I do speak for MSG.Net -- http://msg.net/

Your own domain name/e-mail for only $49/year, no setup fee: dom...@msg.net