The weirdest thing just happened.  I've been running a "make release" on
a sparc, with the sources on an i386 server, both running NetBSD of
course (and in fact 2001/06/19 -current).  Suddenly in the middle of the
'mkdep' for the last kernel of the sparc build it spit up:

	cpp: /proven/work/woods/NetBSD-src/sys/sys/types.h: Input/output error
	cpp: machine/bswap.h: Input/output error
	cpp: /proven/work/woods/NetBSD-src/sys/arch/sparc/sparc/vm_machdep.c: Permission denied
	cpp: /proven/work/woods/NetBSD-src/sys/arch/sparc/sparc/disksubr.c: Permission denied
	cpp: /proven/work/woods/NetBSD-src/sys/arch/sparc/dev/md_root.c: Permission denied
	cpp: /proven/work/woods/NetBSD-src/sys/sys/param.h: Input/output error
	mkdep: compile failed.
	*** Error code 1
	
	Stop.
	make: stopped in /build/NetBSD-obj/arch/sparc/compile/INSTALL
	*** Error code 1

This is after successfully building the GENERIC, GENERIC_SCSI3, and
GENERIC_SUN4U kernels!  Note the "make release" was running as root too.

Curiously it happened at almost exactly the same time I was doing a
"make release" on the sources server and as it was starting to build the
floppy images.

The timestamps on these log entries correspond almost exactly to the
timestamp on my prompt after the make on the sparc died:

	Jun 26 19:18:44 proven /netbsd: vnd0: no disk label
	Jun 26 19:18:45 proven last message repeated 2 times

and the server had just finished building the ramdisk-big.fs image when
I noticed the sparc build had died.

There are no other related log entries on either machine.

All of the files mentioned above are now readable without problem on the
sparc.

Is it possible that the vnd device is somehow having adverse affects on
the NFS server?

Should I send-pr this?

Do I dare try to repeat it?  ;-)

-- 
							Greg A. Woods

+1 416 218-0098      VE3TCP      <gwoods@acm.org>     <woods@robohack.ca>
Planix, Inc. <woods@planix.com>;   Secrets of the Weird <woods@weird.com>