Re: http protocol and a unknown image type causes a reboot

> And that goes into a loop? (reboot, http fetch, reboot, http fetch, reboot)

Yes, the reboot is physical, not a call to the bootstrap again, and it
happens always the same way, so it is a loop, yes.

>    MMmmmmm    "unknown image"
> 
> Could you elaborate "unknown" image? What qualifies an image working
> with TFTP into "unknown" when using HTTP?

Unknown image is an image of a type not know for etherboot, for example a
tar.gz or even a nbi if you don't have nbi support compiled in.

The images don't work in http or tftp, but tftp does the correct thing,
which is printing the "Unable to load file." message and retrying the load
again, on http it doesn't print this message, even though it prints the
"error: not a valid image" one, and then the machine reboots.

> > I'm using gcc 3.3 to compile my images in a Debian environment in case that
> > matters.
> 
> I assume you refereer to Etherboot image now. ( Where you previous were
> refering to payload images ( pxelinux.0 or kernel images )
> Please confirm (or deny ;-)

Yes, sorry for not being more explicit.

> I willing to analyze a sniffed session stored in libpcap format (
> tcpdump / ethereal ).  Put it on a server and post the URL to the ML.

I think that if you try to compile etherboot with http support and without
pxe support and tell it to boot pxelinux.0, for example (I know this is not
going to work), you'll get the reboot, if you cannot get it, just tell me
and I'll do the capture.

Regards...
-- 
Manty/BestiaTester -> http://manty.net


-------------------------------------------------------
SF.Net email is Sponsored by the Better Software Conference & EXPO
September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices
Agile & Plan-Driven Development * Managing Projects & Teams * Testing & QA
Security * Process Improvement & Measurement * http://www.sqe.com/bsce5sf