"Matt D. Robinson" wrote:
In all cases I used the module panic .... The lcrash using vmdump.1 from the
module panic worked well !!
I used the identical panic for vmdump.2 ...... The vmdump.2 size was identical
to the vmdump.1 size however lcrash thought only the headers were dumped !!
I will go test NMI now ....
> Ken Pearlstein wrote:
> >
> > Here is my /etc/sysconfig/vmdump ....
> > DUMP_ACTIVE=1
> > DUMPDEV=/dev/vmdump
> > DUMPDIR=/var/log/vmdump
> > DUMP_SAVE=1
> > DUMP_LEVEL=4
> > DUMP_COMPRESS_PAGES=0
> > PANIC_TIMEOUT=60
> >
> > The first question is )
> > when I force a panic of the system ( I used a module which calls panic)
> > sometimes I get the message "Warning: vmdump only contains a dump
> > header. Printing dump header:" and sometimes I dont ...
>
> You can't use a module to call panic(). lcrash isn't capable (yet)
> of debugging kernel modules. The resulting crash dump may not be
> usable by lcrash. The memory will be there, of course ...
>
> > The vmdump."BOUNDS" files are the same size .....
> > -rw-r--r-- 1 root root 60344741 Mar 13 10:50 vmdump.2
> > -rw-r--r-- 1 root root 60344741 Mar 13 06:54 vmdump.1
> >
> > vmdump.1 works and vmdump.2 doesnt ????
>
> Is the second one an exception dump, and not a panic() dump?
>
> > The second question is) do you know if an NMI will produce a dump ??
>
> It _should_ ... I put in some code to catch the NMI watchdog, but I
> haven't reproduced the case as of yet.
>
> > and the last question) .... when $? is non zero why do the following
> > lines in save_vmdump
> >
> > $LCRASH -r map.$BOUNDS vmdump.$BOUNDS kerntypes.$BOUNDS
> > \
> > > analysis.$BOUNDS 2>&1
> > if [ $? -ne 0 ] ; then
> > echo "Error: could not create crash report." >&2
> >
> > exit 1
> > fi
> >
> > produce the following output
> >
> > Saving crash dump data (if any)
> > Error: could not create crash reAdding Swap: 530112k swap-space
> > (priority -1)
> > port.ut to the sniaconsole
> >
> > I am not sure if this happens on all linux consoles or is only an snia
> > console problem .....
>
> This might be a console issue on your end. Looks like the kernel log
> buffer (dmesg) and the boot-up script information is going to the same
> screen.
>
> --Matt
>
> > --
> > ==========================================================================
> > Ken Pearlstein Field Tech Support Analyst
> >
> > cell phone 703-362-9793 pager 888 308-0734
> > C/O SILICON GRAPHICS pearlste@xxxxxxxxxxxxxxxxxxx
> > 14160 NEWBROOK DRIVE, SUITE 100 voice: (703) 227-8531 430-5403
> > CHANTILLY, VA 20151 fax: (703) 277-8500 430-5403
> > =========================================================================
--
==========================================================================
Ken Pearlstein Field Tech Support Analyst
cell phone 703-362-9793 pager 888 308-0734
C/O SILICON GRAPHICS pearlste@xxxxxxxxxxxxxxxxxxx
14160 NEWBROOK DRIVE, SUITE 100 voice: (703) 227-8531 430-5403
CHANTILLY, VA 20151 fax: (703) 277-8500 430-5403
=========================================================================
|