home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.unix.aix
- Path: sparky!uunet!shared!johnf
- From: johnf@shared.com (John Fusselman)
- Subject: moving files from disk to tape - problems!
- Summary: attempts to move files from disk to tape with dd fail instantly and
- Message-ID: <1992Nov17.163014.66377@shared.com>
- Date: Tue, 17 Nov 92 16:30:14 GMT
- Distribution: usa
- Organization: Shared Financial Systems
- Keywords: dd cp delete tape devices
- Lines: 77
-
- In the process of attempting to duplicate a tape I effortlessly read the
- tape contents onto disk using dd - ``dd if=/dev/rmt0.1 of=FILENAME''. The
- trouble came when I attempted to copy the files from disk to a new tape;
- dd never moved the tape but instead *promptly* produced an error as
- below:
-
- C:9$ dd if=file1 of=/dev/rmt0
- dd write error: Invalid argument
- 17+0 records in
- 16+0 records out
- C:10$
-
- The file, file1, is about 1.5 Mbytes and registered 3102 blocks on the
- way in, also using the default block size; disregard the record count
- shown in the example above because the numbers vary from 3 or more to
- less than 20 from one time to the next. Sometimes they even match,
- indicating the same number of blocks out as in! The command fails so
- quickly it is as though it is indeed complaining about the arguments
- presented (or not presented). A search of info proved unhelpful, only
- confirming my expectations that what I am trying to do is legitimate.
- BTW similar operations under SunOS worked flawlessly over many years.
-
- Either an RS/6000 550 running AIX 3.1 or a 220 running 3.2.2 acts the
- same; the tape drives are both 1/4 inch cartridge, QIC-150 type,
- genuine IBM. The tape drives on both machines appear to function
- normally in all other respects. Not that it should have made a
- difference, but I tried the arguments to dd in different orders and
- with and without a blocksize as well as with "conv=sync." That
- particular argument, according to info, would have an effect only on
- the last block transferred because "streaming tape devices permit only
- multiples of 512 bytes." It's the first blocks that concern me - the
- tape never moves!
-
- What is wrong with the use of dd to copy a file from disk to tape? Has
- dd ever been fully functional under AIX?
-
- Things got more interesting when I tried "cp" on the 220 under 3.2.2;
- sometimes when it failed, it deleted the tape device! I was only able
- consistently to delete rmt0, rmt0.1, .2, and .4; rmt0.5 usually works
- flawlessly - if it works, it will move all of the file to tape, which
- can then be retrieved and compared (with "cmp") and verified to be
- perfectly reproduced and recoverable from the tape. If "cp" deletes
- the device, it will do so on the first try. For what it's worth, the
- "retension on open" devices will do so, even though they disappear.
-
- Has anyone else seen the like of this, the removal of a device upon an
- error from cp? The response from cp was as follows:
-
- M:# cp file1 /dev/rmt0.1
- cp: /dev/rmt0.1: A system call received a parameter that is not valid.
- cp: /dev/rmt0.1: A system call received a parameter that is not valid.
- M:#
-
- Thereafter, the device, rmt0.1, would be gone.
-
- I was logged in as root at the time the devices were deleted, but
- neither as root nor under my usual login was I able to move a tape with
- dd with an output file ("of=/dev/rmt0") of a tape device. Tape to disk
- and disk to disk operations - whether to or from remote machines - have
- always performed as expected.
-
- Calling IBM on the problem of the disappearing tape devices yielded the
- response "that you're trying to do something unorthodox, using cp to
- copy a file to tape; the result should be that the one file replaces
- the other." But I said it did NOT replace it; it blew it away! "No
- problem; the devices are rebuilt when the system is rebooted!" I will
- NOT dignify this response with a comment of my own.
-
- At this time I await the arrival - sometime this week or next, I
- presume - of my local SE to "see what I'm doing wrong or to determine
- what's wrong with my systems." IMHO I think what's wrong is the
- nameplate on the front and the OS it's running!
-
- Thanks for sharing your experiences and knowledge.
- --
- Dallas, Texas "Visualize whirled peas."
- --->>> My comments are my opinions and may not be shared by Shared <<<---
-