parameters for (r)dump to an exabyte
Ken Trant
trant at empower.corp.sgi.com
Wed Jun 26 09:48:24 AEST 1991
The parameter used (6000) is wrong. The way I read his command line is he wants
a level 0 dump, write the successful completion to /etc/dumpdates, tape length of
6000 feet, density and dump device.
The exebyte is 112 meters or 4409.44 inches, or 453572 bpi (2000Mbytes on American
exabyte tapes). I use 408215 as a density (minus 10%) for saftey sake.
Also remember to use the no rewind device for multiple volumes on 1 tape.
Good luck
In article <frazier.677435824 at oahu>, frazier at oahu.cs.ucla.edu (Greg Frazier) writes:
|> janet at cs.uwa.oz.au (Janet Jackson) writes:
|>
|> >In <1991Jun19.222701.29689 at cs.ucla.edu> frazier at oahu.cs.ucla.edu (Greg Frazier) writes:
|>
|> >>Hello -
|> >> Could somebody give me the appropriate parameters for
|> >>doing a dump to an exebyte tape (2 gbyte)? For the record,
|> >>what I just tried was
|> >> dump 0usdf 6000 54000 /dev/rst1 filesystem
|> >>as specified in the man page. Thanks for any help.
|>
|> >And what went wrong?
|>
|> >I'm using the same parameters, except I'm using a blocksize of 126, and it
|> >seems to work fine. (Don't think I've ever reached the end of a tape yet,
|> >though :-{ ) (that's supposed to be a look of consternation)
|>
|> For the curious, here is the context. I am writing a dump script,
|> and was testing/debugging it. An error msg similar to this appeared
|> on the console:
|>
|> error on /dev/rst1: bad length parameter
|>
|> Let me say now that I'm a novice at system hacking. I did
|> a (mt -f /dev/rst1 fsf 1) to skip a 10 meg file at the front
|> of the tape, and attempted to dump 2 file systems onto the tape.
|> These should have come to much less than 1 gig, let alone 2.3 gigs.
|> I was not planning on keeping track of the amnt of tape already
|> used, but rather to guarentee that the file systems being dumped
|> on a single tape together had a lower capacity than the tape. It
|> has been suggested to me that rdump'ing w/out specifying a blocksize
|> is bad news - I'll try that as my first correction.
|> --
|>
|>
|> Greg Frazier frazier at CS.UCLA.EDU !{ucbvax,rutgers}!ucla-cs!frazier
--
Ken Trant <trant at sgi.com> / Third Star to the right
Information Services, / And straight on till
Silicon Graphics, Inc / Morning
More information about the Comp.unix.admin
mailing list