[EMAIL PROTECTED] (Tim Jones) writes:
> Ben Brillat wrote:
> > Do you mean that the Ditto Max WON'T work with linux unless I have a P200,
> > or it will work SLOWLY...??
> >
> > Tim Jones wrote:
> > > Bruce - see my last comment to Ernst. No joy with Max on less than a
> > > P5-166MHz. In fact, the Max box even states - "Pentiuim 150 required
> > > for maximum performance." This goes DOUBLE for Linux.
>
> It will sometimes work slowly (<120K per sec), but mostly you will
> receive ECC and CRC errors caused by too slow data input to the drive.
> I would not trust data written to the tapes under those conditions -
> even with BRU (there's only SO much we can fix)!
I find I am not receiving ECC and CRC errors on a P133 (which is faster
than the minimum spec listed in the system requirements on the Max box).
In addition, the number of repositions is low.
There are a few things that I've done help this.
The first is to limit the data rate to 2000 (in /etc/conf.modules).
Otherwise, ftape starts at the 4000Mb/s the accellerator can support,
errors are reported and it backs off. Sometimes it only went down
to 3000, but errors were reported, and the fifo threshold had to be
increased. So, I stick to 2000Mb/s.
alias char-major-27 zftape
options ftape ft_fdc_driver=ftape-internal,none,none,none # ft_tracings=4
options ftape-internal ft_fdc_base=0x370 ft_fdc_irq=6 ft_fdc_dma=3
ft_fdc_rate_limit=2000
pre-install ftape-internal /sbin/swapout 16
Secondly, I found that enabling multi-sector transfer and interrupt
unmasking for my IDE disk stopped the number of times the tape was
repositioned during a backup. The usual warnings about using interrupt
unmasking apply -- see hdparm manual page.
hdparm -u 1 -m 16 /dev/hda
hdparm -u 1 -m 16 /dev/hdb
Finally, I leave the machine otherwise idle when it is performing
the backup.
As I remember it, the data rate reported by dump is ~160K/s.
--
`O O' | Home: [EMAIL PROTECTED] http://www.alfie.demon.co.uk/
// ^ \\ | Work: [EMAIL PROTECTED]
