home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!olivea!sgigate!odin!fido!zola!zuni!anchor!olson
- From: olson@anchor.esd.sgi.com (Dave Olson)
- Newsgroups: comp.unix.admin
- Subject: Re: multiple dumps on EXB-8200 8mm tape
- Message-ID: <noca1fk@zuni.esd.sgi.com>
- Date: 25 Jul 92 18:22:58 GMT
- References: <1992Jul23.141833.14606@ferrari.nmc.ed.ray.com> <1992Jul24.181553.10455@kakwa.ucs.ualberta.ca>
- Sender: news@zuni.esd.sgi.com (Net News)
- Distribution: usa
- Organization: Silicon Graphics, Inc. Mountain View, CA
- Lines: 56
-
- In <1992Jul24.181553.10455@kakwa.ucs.ualberta.ca> sherwood@fenris.space.ualberta.ca (Sherwood Botsford) writes:
-
- | Bill Heiser writes
- | > The problem I have is that when I write a tapefile to the non-rewinding
- | > device, and then write another file, when I try to read the files I seem
- | > to have to skip filemarks sometimes. For example: If I write a tape on
- | > a SunOS 4.1.1_U1 or 4.0.3 system, I have a tape with several tapefiles.
- | > If I try to read the tape by doing several consecutive 'restore' commands
- | > on the non-rewinding device, the restores after the first one fail -- unless
- | > I do a restore, then a "mt -f fsf 1". So basically I seem to be getting
- | > extra file marks!
- |
- | This one threw me for a loop initially.
- | The file mark is not (usually) explicitly written by the program, but is
- | written by the computer on closing the file. It will aslo be done by the tape
- | if the previous command was write data, and the the current command is any tape
- | movement command.
-
- This is incorrect. The drive itself will *never* write a filemark. It
- will write an EOD mark if you are writing, then rewind, space backwards,
- etc.
-
- | Now, a filemark on either exabyte or DAT drives has a front and back side.
- | Front being on the Beginning of Tape side.
- |
- | Any command that writes to the tape, finishes with a write filemark, and
- | leaves the head on the back side of the mark, ready to write the next file.
- |
- | Any command that reads the tape leaves the tape positioned on the BOT side of
- | the file mark.
-
- Wrong. A read that actually encounters the FM (e.g., a read when
- positioned at the BOT side of a FM) will return no data (or short
- count, if you had some data before you hit the FM), and the tape
- will be positioned on the EOT side of the FM. Exabyte has a nice
- series of diagrams in their appendix covering just about every
- possible case for FM's, EOD, etc. that explain this quite well.
-
- | mt fsf's and bsf's behave this way too. A bsf leaves the tape on the front
- | edge of the filemakr, and an fsf leaves it on the back.
- |
- | So if you kill an operation, and want to reposition the tape to start it again,
- | you must do this:
- | mt bsf 1
- | mt fsf 1
-
- This is wrong. If you want to record on an Exabyte 8200, you
- must be in one of 3 places: BOT, BOT side of a FM, or EOD. If I
- remember correctly, the 8500 will also allow you to record at the EOD
- side of a FM. So you do not want to do the mt fsf 1 after the bsf
- if you want to overwrite some data on the tape for 8200's
-
- --
- Let no one tell me that silence gives consent, | Dave Olson
- because whoever is silent dissents. | Silicon Graphics, Inc.
- Maria Isabel Barreno | olson@sgi.com
-