On Tue, Oct 26, 2004 at 03:45:16PM -0600, John E Hein wrote:
> Joe Rhett wrote at 14:28 -0700 on Oct 26, 2004:
> > On Mon, Oct 25, 2004 at 06:35:44PM -0400, Gene Heskett wrote:
> > > If the file cannot be restarted from byte 1 on the next tape, eg the
> > > system cannot back up to the start of the file being written, then
> > > something is seriously wrong with the method being used. In the case
> > > of no holding disk, then it seems setting one up would be the answer.
> >
> > These are all stored on a holding disk, and I see the same problem during
> > amflush of that holding disk.
> >
> > What's even more curious is that the holding disk files are broken up into
> > 1 gigabyte segments. So 30-something of these would fit on a single tape,
> > and if EOT was hit before tape-end, then only <1gb would be lost.
> >
> > Instead, nothing is being recorded as written to the tape.
>
> What is the setting for 'taperalgo' in your config?
It's default, which is 'first'.
I could try 'largestfit', but nothing in that flush is actually larger than
5gb so I'm not sure that this applies to amflush...
--
Joe Rhett
Senior Geek
Meer.net
|