[comp.unix.ultrix] Multivolume dumps via dump | compress | dd

jeff@nic.the.net (Jeff Hayward) (03/01/90)

I'm trying to save some space and time by compressing my dumps before writing
them to tk50 on my DS3100 running UWS 2.1.  I use 

	dump 0ucf - /usr | compress | dd of=/dev/rmt0h obs=60k

to write a dump, which seems to work fine. dd prompts me to replace cartridges
when appropriate, and dump seems happy. 

A restore of the form

	dd if=/dev/rmt0h ibs=60k | uncompress | restore if -

works, until the end of the first tape is reached.  I then get "read: i/o
error" from restore (I think).  dd just tells me its usual records in/out
count. I've tried using the files=n argument to dd, to no avail.  Can anyone
shed light on what I'm doing wrong?

Jeff Hayward
The University of Texas System		jah@mojo.ots.utexas.edu
Office of Telecommunication Services	+1 512 471 2427

avolio@decuac.DEC.COM (Frederick M. Avolio) (03/01/90)

I found the following command saves the most space for dumps and I only ever
need mount one tape:

dump 0ucf - /usr >/dev/null

Of course this ius useful only if you don't ever need to restore :-)

Fred


------
Look, just in case, the above is a joke, see?  Don't go saving your disks
like that!

grr@cbmvax.commodore.com (George Robbins) (03/01/90)

In article <25276@ut-emx.UUCP> jeff@nic.the.net writes:
> I'm trying to save some space and time by compressing my dumps before writing
> them to tk50 on my DS3100 running UWS 2.1.  I use 
> 
> 	dump 0ucf - /usr | compress | dd of=/dev/rmt0h obs=60k
> 
> to write a dump, which seems to work fine. dd prompts me to replace cartridges
> when appropriate, and dump seems happy. 
> 
> A restore of the form
> 
> 	dd if=/dev/rmt0h ibs=60k | uncompress | restore if -
> 
> works, until the end of the first tape is reached.  I then get "read: i/o
> error" from restore (I think).  dd just tells me its usual records in/out
> count. I've tried using the files=n argument to dd, to no avail.  Can anyone
> shed light on what I'm doing wrong?

Try something like:

(dd if=/dev/rmt0h ibs=60k; mt offline; \
	echo "change tape, hit return" > /dev/tty ; line > /dev/null ; \
	dd if=/dev/rmt0h ibs=60k) | \
    uncompress | restore if -

The trick would be to make the input to uncompress|restore look "seamless"
while actually giving you a chance to change the tape without getting an
error trying to read the tape.

Uh, no, I didn't try it...
-- 
George Robbins - now working for,     uucp:   {uunet|pyramid|rutgers}!cbmvax!grr
but no way officially representing:   domain: grr@cbmvax.commodore.com
Commodore, Engineering Department     phone:  215-431-9349 (only by moonlite)

cmaarpc@cc.ic.ac.uk (Peter Churchyard) (03/02/90)

Fred, use alias dump echo Dump Done.
Pete.
p.s. See you all in New Orleans!

litwack@dccs.upenn.edu (Mark Litwack) (03/20/90)

In article <25276@ut-emx.UUCP> jeff@nic.the.net writes:
: I'm trying to save some space and time by compressing my dumps before writing
: them to tk50 on my DS3100 running UWS 2.1.  I use
:
:       dump 0ucf - /usr | compress | dd of=/dev/rmt0h obs=60k
:
: to write a dump, which seems to work fine. dd prompts me to replace
: cartridges
: when appropriate, and dump seems happy.
:
: A restore of the form
:
:       dd if=/dev/rmt0h ibs=60k | uncompress | restore if -
:
: works, until the end of the first tape is reached.


It seems to me that using compress for doing backups is not a good idea.
If you get a bad spot on your tape and cannot read it, the rest of your tape
will be unreadable since uncompress cannot continue uncompressing with
missing blocks in the middle of the file.

Restore by itself will at least attempt to re-sync itself if it encounters
bad areas on the tape.

Dump really needs to do a compress for each file it saves.

-mark