X-Git-Url: https://git.gag.com/?a=blobdiff_plain;f=docs%2Fzftape.txt;h=e69de29bb2d1d6434b8b29ae775ad8c2e48c5391;hb=d92f70685083588e2a7ce6bc312a735f6937b5a6;hp=7d84ce665dc9aa83fec97c47b61545ab18414e98;hpb=6c1f39091444e58c33362f0cc086375d9d273e77;p=debian%2Famanda diff --git a/docs/zftape.txt b/docs/zftape.txt index 7d84ce6..e69de29 100644 --- a/docs/zftape.txt +++ b/docs/zftape.txt @@ -1,109 +0,0 @@ - -Chapter 35. Usage of floppy tape drives on Linux -Prev Part VI. Historical files Next - -------------------------------------------------------------------------------- - -Chapter 35. Usage of floppy tape drives on Linux - - -Albrecht Gebhardt - -Original text - -Stefan G. Weichinger - -XML-conversion,Updates -AMANDA Core Team - -Amanda now supports the ftape driver version 3.04d (see http://www- -math.math.rwth-aachen.de/~LBFM/claus/ftape ). It adjusts the blocksize -automatically to 32k and supports QIC volume tables. -It uses only one open() call for writing backups to one tape, so the "busy"- -lamp of the drive will be on all the time. (With normal Amanda code it would be -off in pauses between two files written to tape.) -For volume table support you have to get libvtblc first (available at ftp:// -pc02-stat.sci.uni-klu.ac.at/pub/Linux/libvtblc.) This library contains the -functions and subroutines from the vtblc utility, distributed with ftape. -(Maybe this library will be part of a future ftape package). -You have to set the raw tape device for volume table operations, usually /dev/ -rawft0, either via configure --with-ftape-rawdevice= or with rawtapedev in -amanda.conf (configure checks for /dev/rawft[0-3], to get a guess for this -value). Dont forget to make this device read/writeable for your backup user. -For compilation you need the header files from ftape 3.04d in your include -tree. -The volumetable of a tape "amlabeled" TEST-VOL2 with 4 backups on it would look -like (listed with vtblc utility from ftape): - - prompt: vtblc - - Nr Id Label Date Start End - Space - - 0 VTBL TEST-VOL2 23:08:06 03/15/98 3 4 0.00 - % - 1 VTBL gamma //beta/C 0 00:00:00 03/15/98 5 374 0.68 - % - 2 VTBL gamma sda2 0 00:00:00 03/15/98 375 1029 1.21 - % - 3 VTBL alpha sda2 0 00:00:00 03/15/98 1030 1906 1.62 - % - 4 VTBL alpha sda6 0 00:00:00 03/15/98 1907 8092 11.45 - % - 5 VTBL Amanda Tape End 01:45:15 03/16/98 8093 8094 0.00 - % - -With lvtblc, currently available with the libvtblc library, you can list the -complete label strings (44 characters, not only the first 22 characters as with -vtblc): - - - prompt: lvtblc -l - - Nr Id Label Date - - 0 VTBL TEST-VOL2 23:08:06 03/15/98 - 1 VTBL gamma //beta/C 0 00:00:00 03/15/98 - 2 VTBL gamma sda2 0 00:00:00 03/15/98 - 3 VTBL alpha sda2 0 00:00:00 03/15/98 - 4 VTBL alpha sda6 0 00:00:00 03/15/98 - 5 VTBL Amanda Tape End 01:45:15 03/16/98 - -Note on datestamps: volume 0 (Amanda label): reflects the time of starting the -backup volume i (backup files): Amanda datestamps of the backup files last -volume (end marker) : reflects the time of finishing the backup -I tested this on a Linux machine (P90 / 96Mb RAM / 256 Mb swap) with two other -clients (Linux / WfW) using - -* Linux Kernel 2.0.33, -* ftape 3.04d, -* Amanda 2.4.0b6p4 - -with an internal Iomega Ditto 3200 (TR-3) drive attached to an Iomega Ditto -Dash controller (at 2000 Kbps). My tapetype follows: - - define tapetype DITTO-TR3 { - comment "Iomega DITTO 3200 Travan 3 tape drives" - length 1500 mbytes # - filemark 29 kbytes # ??? - speed 256 kbytes # = 2000 Kbit/s ? - } - - - -Note - -Filemarks are not written to the tape (they are written to the header segment -and use there 128 byte), so their size could even be 0. But a tape segment -takes at least 29 kb (+3 kb ecc-code = 32 kb), so in the worst case an eof will -waste 29 kb. - -Note - -Refer to http://www.amanda.org/docs/zftape.html for the current version of this -document. -------------------------------------------------------------------------------- - -Prev Up Next -Chapter 34. Y2K Compliancy Home Part VII. Appendixes -