On Mon, 8 Oct 2001 at 5:36pm, Jeremy Wadsack wrote
> Lee Fedor ([EMAIL PROTECTED]):
>
> >> define tapetype EXB-8505XL {
> >> comment "Exabyte 8505XL tapes (works with Exabyte 10h changer)"
> >> length 9584 mbytes
> >> filemark 917280 kbytes
> >> speed 667 kps
> >> }
> >>
> >> 1] I'm fairly certain these are 7GB tapes (and even gave -e 7g to
> >> tapetype when I ran it). Why does tapetype come back saying they are
> >> 9.5GB?
>
> > it getting 9.5 with software compression. which is reasonable.
Tapetype doesn't use software compression. It just writes random data to
the drive until it hits EOT. This also means that if harware compression
is enabled, you'll actually see a much *smaller* length than you expect,
since random data doesn't compress very well. Here's what I got running
tape type on an Eliant 820 with XL tapes (the Eliant is the same size as
the 8505XL, just faster):
define tapetype Eliant-820 {
comment "just produced by tapetype program"
length 6680 mbytes
filemark 0 kbytes
speed 949 kps
}
> Hmm.. That make sense. I guess I need to figure out how to turn off
> hardware compressions the, since software compression should be better
> and faster.
Better, yes -- faster, that depends on your CPU speed. The advantage of
software compression is that amanda will be able to plan better, since
she'll keep a history of how well each fs compresses, and so knows how big
each partition when it gets to the tape. With hardware compression, you
have to guess how well your filesystems will compress, and increase the
length tapetype reported accordingly.
> Why not have Amanda automatically rewind the tape when it's done. Or
> when it starts? That just seems to make sense since all the operations
> would expect the tape to be rewound.
Amanda *never* appends, and does rewind the tape when she starts up (but
not when she's done). Do you mean that in a single night's run, using
multiple tapes, you're only getting 2GB per tape? That *could* be a
result of combined software/hardware compression, but it seems rather
extreme. Are there any error messages in the system logs?
--
Joshua Baker-LePain
Department of Biomedical Engineering
Duke University
