Re: Deadlock situation?

On Thu, May 1, 2008 at 9:38 AM, Dan Armbrust
<daniel(dot)armbrust(dot)list(at)gmail(dot)com> wrote:
> On Wed, Apr 30, 2008 at 12:36 PM, <dan(at)sidhe(dot)org> wrote:
> > > In this case, Postgres had been started in the foreground on a
> > > terminal, so I went to that terminal, and did a ctrl-c. Eventually,
> > > postgres stopped, but the terminal wouldn't respond either - and I had
> > > to close it.
> >
> > Just out of curiosity, could you maybe have XOFF'd the terminal? I've had
> > cases in the past where Postgres got unhappy and froze when it couldn't
> > write logging information to stderr. (Though, granted, in the cases I hit
> > the server's stderr was redirected to a pipe that was full, so it's not
> > 100% analogous)
>
> Its certainly possible that something "interesting" happened to the
> terminal. We do have an issue with a KVM switch on this machine that
> sometimes causes all sorts of unpredictable random garbage input to
> hit the screen when you switch to the system.

Place I worked at 5 to 10 years or so ago, I specifically did NOT
allow them to hook up the KVM switch to my primary servers for just
this reason, plus it kept the junior super administrators from doing
stupid things like hitting CTRL-ALT-DELETE on one of my servers. On
the very very rare occasion I needed to actually sit in front of the
server I'd hook up my one kvm connector and do what I had to, then I'd
disconnect it and do everything else remotely.

Admittedly, it wasn't the KVM that was at fault there, but it sure did
cause some heart ache when someone sat down and hit CTRL-ALT-DEL to
log into windows without looking at the screen and seeing that the
last machine it was on was a Red Hat box...