nbd swap server, inetd, SO_KEEPALIVE

[Gavin McCullagh]

Hi,

On Tue, 08 May 2007, Gavin McCullagh wrote:

> > Adding the following line to /etc/hosts.allow:
> > 
> > nbdswapd: ALL: keepalive
> 
> Excellent.  I've done the same, though changing ALL to my local subnet
> only.

I applied this fix yesterday.  Since then, I'm regularly getting this
(previously unseen) message in the logs:

  May  9 13:34:43 medlycott nbd_server[28314]: Read failed: Connection reset by peer

which I guess is probably consistent with tcpd timing out a dead connection
and resetting and nbd_server just getting reset.

Two questions:

1. Do people think this inference is likely correct or is it a concidence and
   something is really wrong?

2. I guess I should report a minor bug to logcheck guys to go along with
   the keepalive option being added in /etc/hosts.allow so that this
   message, if it's really normal, gets ignored by logcheck.

Gavin