On Tue, 12 Jun 2001, Russel Ingram wrote:
> On Wed, 13 Jun 2001, Ivan Rayner wrote:
>
> > Well, even if it is a hardware problem, it'd be nice if xfsdump didn't
> > core dump. Would you be able to send me the output of 'xfsdump -v5 ...'
> > and the core file?
>
> I can't seem to get the error messages to redirect to the file so here are
> the error messages I got this time:
>
> sh: xfsdq: command not found
> sh: xfsdq: command not found
These errors indicate that xfsdq/xfsrq weren't installed. If you want
your quota information included in the dump, you should install them. I
think they're part of the xfsdump package. If you don't care, then you
can ignore these errors.
> xfsdump: drive_minrmt.c:2201: do_end_write: Assertion `first_rec_w_err >=
> 0' failed.
> Aborted (core dumped)
It looks as though the following code failed with errno == 22 (Invalid
argument):
cmd/xfsdump/common/drive_minrmt.c:
rval = ioctl( fd, MTIOCTOP, &mop );
if ( rval < 0 ) {
/* failure
*/
mlog( MLOG_DEBUG | MLOG_DRIVE,
"tape op %s %d returns %d: errno == %d (%s)\n",
printstr,
param,
rval,
errno,
strerror( errno ));
return -1;
}
where mop.mt_op == MTWEOF and mop.mt_count == 1, which later triggered
the assertion failure and core dump.
In other words, your hardware/driver didn't want to write an
end-of-file record. You could try running 'mt weof 1' from the command
line to verify my assertion.
What driver does your tape drive use? You could possibly take a look
at the source and see how it handles the MTIOCTOP (MTWEOF) ioctl. For
example, you can see the scsi tape driver code for this at
linux/drivers/scsi/st.c line 2223.
Ivan
--
Ivan Rayner
ivanr@xxxxxxxxxxxxxxxxx
|