[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: Problem with inetd and pop3d looping

> I am running RH5.0 with 2.0.34 kernel. Recently we have seen an increase
> in the number of failures of ipop3. The message below was in the
> messages file. 
> Aug 24 11:25:46 ns1 inetd[561]: pop-3/tcp server failing (looping),
> service terminated 
> It seems to happen about once an hour under heavy load. Does anyone have
> any ideas as to why this started occuring? The problem occurs running a
> stock 2.0.30 kernel.

This happened once to one of our customers on an Intel box.
It is the result of an excessive number of POP3 connections per minute:
inetd turns off the service as a safety measure, assuming
that it could be an attack.
You may increase the max allowed no of connections per minute
in inetd.conf in this way (for instance):
pop-3   stream  tcp     nowait.180  root    /usr/sbin/tcpd      ipop3d
(to get 180 connections per minute instead of the default, 40?)

But if you do not believe that your box really has that many POP connections,
check for clock modifications: the threshold calculation may give
spurious results if the system date is changed abruptly with the
date command.  This is what actually happened to our customer
which was happily playing with the date command on his production
system :-)

furio ercolessi
Spin - Trieste (Italy)


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index] []