Re: Very wierd behavior on new nodes
On Tue, 2003-01-07 at 09:25, Arthur H.
Edwards,1,505-853-6042,505-256-0834 wrote:
> Under the 2.4.19 kernel, it depends on whether the job was spauned using
> mpi. If I start it on the node, it does not kill communications. If I
> start it from a head node, I can still ping. I have only tried the 2.2
So it doesn't kill the box, but makes it effectively unusable?
If you have a remote telnet/ssh session already open to the box, is it
knocked out? I'd open that remote session and run top(1) on it to try
glean a little more info. If you *do* get kicked off, that certainly
would tell us something!!!!
> kernel job as a stand-alone. There it exits quite gracefully with exit
> 139. Do you know what exit 139 is?
Nope, but it looks like Colin does...
> Art Edwards
>
> Ron Johnson wrote:
>
> >On Mon, 2003-01-06 at 22:48, Arthur H.
> >Edwards,1,505-853-6042,505-256-0834 wrote:
> >
> >
> >>I'm having monumental difficulty getting a new set of PC's working. I
> >>had been installing a 2.4.19 kernel with debian on a MB with a via chip
> >>set, and athlon XP2100, a promise ide system. Debian semms to install
> >>correctly. However, when running large fortran jobs (under g77-3.2), the
> >>system would either die immedieately, or start running and then die.
> >>When I say die I mean that I can't login. I have backed off to a 2.2.20
> >>kernel and g77 2.95. Now the program dies with an exit 139, but the
> >>system stays up.
> >>
> >>
> >
> >Can't login because your fortran program is taking too much CPU?
> >
> >Can you still ping the box from another node?
> >
--
+------------------------------------------------------------+
| Ron Johnson, Jr. mailto:ron.l.johnson@cox.net |
| Jefferson, LA USA http://members.cox.net/ron.l.johnson |
| |
| "Basically, I got on the plane with a bomb. Basically, I |
| tried to ignite it. Basically, yeah, I intended to damage |
| the plane." |
| RICHARD REID, who tried to blow up American Airlines |
| Flight 63 |
+------------------------------------------------------------+
Reply to: