EXB-8[25]00 dump parameters
2007-12-25 7:44:00
and EXB-8500 tape drive.
Kes
- Kes Masalaitis INTERNET: kes@gvl.unisys.com
Paramax Systems (A Unisys Co.) UUCP: uunet!cbmvax!gvls1!kes
70 E. Swedesford Road PHONE: (215)648-2691
Paoli Pa. 19301 All standard disclaimers apply
Thanks to:
From: phil@pex.eecs.nwu.edu (William LeFebvre)
From: Howie Kaye <howie@ivory.cc.columbia.edu>
From: Mike Raffety <miker@sbcoc.com>
From: celeste@stokely.mtview.ca.us (Celeste Stokely)
From: hydres!paul (Paul Humphreys)
From: birger@vest.sdata.no (Birger A. Wathne)
From: mks!andy@watserv1.uwaterloo.ca (Andy Toy)
From: mark@maui.Qualcomm.COM (Mark Erikson)
From: ept@eptsun1.ctd.ornl.gov (E P Tinnel)
From: era@niwot.scd.ucar.EDU (Ed Arnold)
My original question:
>Hi,
>
>I was wondering what people use for dump parameters for the
>EXB-8200 and EXB-8500. I have been using:
>
>rdump 0bdsfu 126 54000 6000 <dumpdest> <filesystem>
>
>for the 8200 but I'm not positive this is correct. Are these parameters
>allowing the full use of the tape? Also, what changes to these parameters
>need to be made to allow for the EXB-8500 tape drive?
>
--------------------------------------------------------------------------
From: phil@pex.eecs.nwu.edu (William LeFebvre)
rdump 0bdsfu 126 54000 6000 <dumpdest> <filesystem>
That's what we use for 8200's. We don't have any 8500's (yet).
From: Howie Kaye <howie@ivory.cc.columbia.edu>
we used to use:
rdump 0ufbsd host:drive 50 6000 54000 filesystem
This was for an 8200. probably doubling the size (but not the blocking
factor) for an 8500 would work.
From: Mike Raffety <miker@sbcoc.com>
You need to double the 6000 parameter for the EXB-8500.
From: celeste@stokely.mtview.ca.us (Celeste Stokely)
Here's a summary I captured from sun-managers on this in April.
Here are the results from my query about dump parameters for an Exabyte 8200.
It is interesting the wide variety of responses (apparently the driver uses
some unknown linear combination of the parameters). I called Exabyte and
they recommended for a 112M tape: 80 54000 6000 and said to use a ratio of
the lengths for the 15M, 54M, etc. They also said that SUNOS has a limit
of 2^31 bytes which is about 2.1G and not 2.3G for a single I/O stream.
Many thanks to all who replied.
Greg
-------------------------------------------------------
Greg Gilley
gilley@ndl.COM [Numerical Design Limited]
919-929-2917 (voice)
-----------------------------------------------------------------------
From: Tom Conroy <trc@NSD.3Com.COM>
Hi Greg:
The only parameter that will change will be 's' for size. This has
absolutely no effect on the data written to tape, only on the amount
of tape that dump estimates that it will use and when it thinks it
has run out of tape.
On a Sony P6-120MP (120M), I use 6000 for a size when matched with
the density and blocking in:
/etc/dump 0udbsf 54000 100 6000 /dev/nrst1 /dev/sd0a
So I would use the following:
Tape Length Size Parameter
120 6000
115 5750
54 2700
15 750
Remember, these are only estimates and come with no warranty whatsoever! :^)
Good Luck!
-----------------------------------------------------------------------
From: ebumfr@ebu.ericsson.se (Michael Rembis 6259)
This is what I use for my Sony P6-120M tapes:
/usr/etc/dump 0ubdsf 56 54000 6000 machine:/dev/nrst9 /dev/xy0h
Chao !
-----------------------------------------------------------------------
From: Eckhard.Rueggeberg@ts.go.dlr.de
We do it with the following script (for 90min tapes) :
if [ $# != 2 ]; then
echo ""
echo " Usage : dump_on_exabyte machine_name exatapeserver_name "
echo ""
exit 1
fi
if [ `ping $1 | grep -c alive` != 1 ]; then
echo $1 lebt nicht !!
exit 1
fi
if [ `ping $1 | grep -c alive` != 1 ]; then
echo $1 lebt nicht !!
exit 1
fi
DUMPFILE=dump.tmp.$$
rsh $1 df | grep dev | colrm 10 > $DUMPFILE
for i
in `cat $DUMPFILE`
do
echo ""
echo Dumping $i from $1 on $2
echo ""
rsh $1 /etc/dump 0ucbfs 126 nobody@$2:/dev/nrst9 20000 $i
done
rm $DUMPFILE
-----------------------------------------------------------------------
From: kkinners@ios.bc.ca (Kerry Kinnersley)
For the 8200 and the 115M tape I use:
0ucbsdf 126 6000 54000 for level 0 dump.
I don't see why you would want to use the 15M and 54M tapes since they are
so cheap. Don't forget the exabyte summaries are in ftphost.uni-augsburg.de
under /pub, perhaps they are listed in there?
Good Luck!
-----------------------------------------------------------------------
From: era@niwot.scd.ucar.EDU (Ed Arnold)
I don't know what you mean by 15M, 54M, and 115M tapes. The usual
dump parms we use with an 8200 writing to a sony P6-120 cartridge are:
dump 0ubdsf 126 54000 6000 /dev/nrst0 /dev/sd2c
-----------------------------------------------------------------------
From: ept@eptsun1.ctd.ornl.gov (E P Tinnel)
Hi ...
try
/etc/dump 0ufbsd /dev/nrst1 100 6000 54000 /dev/rsd0a
for 112m tapes
-----------------------------------------------------------------------
From: herman@galileo.csc.ti.com (Herman Schuurman)
Although I have never used dump on our 8200 (we use Networker for
backups), the documentation suggested the following parameters:
b=50, s=6000, d=54000
I guess you could adapt the size parameter (6000 for a 112m tape), to
account for the smaller 15M and 54M tapes.
-----------------------------------------------------------------------
From: ctm@sleepy.boeing.com (chuck malmsten)
We have an bunch of exebyte 8mm drives. I don't know the model
number. Is an 8200 an 8mm drive.
anyway, we use 8mm P120 tapes which hold about 1.3 gigabytes. The only "strange" thing
we do to get dump to write lots of data to these tapes is to tell dump the
tape density is 54000 bytes per inch. We then specify the tape length
accordingly (e.g. 2000 feet for 1.3 gig).
-----------------------------------------------------------------------
From: oran@spg.amdahl.com (Oran Davis)
Hi,
Dump uses the 9 track standard as the reference for tape length and capacity.
1600bpi 9 track is standard. The effective length that the dump must see is as though
it is writing to 9 track.
the size parameter you pass to dump(8):
------------------- 4.1.1 manual ----------------------------
s size
Specify the size of the volume being dumped to. When
the specified size is reached, dump waits for you to
change the volume. dump interprets the specified size
as the length in feet for tapes, and cartridges and as
the number of 1024 byte blocks for diskettes. The fol-
lowing are defaults:
tape 2300 feet
cartridge 425 feet
diskette 1422 blocks (Corresponds to a 1.44 Mb
diskette, with one cylinder reserved
for bad block information.)
------------------- 4.1.1 manual ----------------------------
G = 1024*1024 (yes! number of 1M chunks)
ts = 2.3*G or 5*G or less if you have short or partially filled tape.
bf = blocking factor. The calture says 124 is best, <64K.
size = ((512 * bf ) + 1920) * (ts - 2048) / (bf * 10667)
example:
103400.49 = (512*124+1920)*(2044*1024-2048)/124/10667
dump 0fbs /dev/nrst0 124 103400 /dev/rsd0g
The variouse tape sizes (ts) and standards (on the NTSC transports we all use):
Minutes NTSC PAL
15 258M 368M
30 516M 753M
60 1.032G 1.484G
90 1.548G 2.226G
120 2.044G (PAL: don't use - too thin and will give errors quickly)
Good luck,
>- Oran
-----------------------------------------------------------------------
From: Vlastimil Malinek <vlastimil.malinek@mrc-apu.cam.ac.uk>
Greg,
After much conversation with sun and other colleagues we use the
following parameters on our switchable 8500 exabytes.
# 2 gig length
#length=5190
# 5 gig length
length=10380
/usr/etc/dump ${level}bcdsuf 112 4100000 ${length}
--------------------------------End of summary from Celeste
From: hydres!paul (Paul Humphreys)
for my exabyter 8200 2.5Gb drive I use /usr/etc/dump 0ucbsdf 56 5190 4100000
Hope this helps. I am getting a 8500 so would be interested in what to use
for it.....
From: birger@vest.sdata.no (Birger A. Wathne)
>From the 4.1.2 man page on dump:
2.3-GByte 8mm tape: dump dsb 54000 6000 126
For 5Gb, use the [n]rst8 and upwards devices, and double the length.
From: mks!andy@watserv1.uwaterloo.ca (Andy Toy)
I use the following /etc/media file on my Solbourne system with an
EXB-8500 using the command:
nice -4 /etc/dump 0uMf P6-120_H /dev/rst9 /
I've included just the Exabyte entries.
#
# TAPE/DUMP MEDIA DATA BASE:
#
# name density passes length irg error ntrec xrate
#
# units bytes/"/ times '->ft inner % 1K Kb/sec
# pass over "->inch record of blocks/
# tape or tape record
# to stream
# EOT gap
# 1/10"
#---- EXABYTE EXB-8200 MEDIA (P6)
X_256 466033 1 48' 0 4 63 250
X_512 466033 1 91' 0 3 63 250
X_1024 466033 1 176' 0 3 63 250
X_1536 466033 1 261' 0 3 63 250
X_2048 466033 1 346' 0 3 63 250
P6-15 466033 1 48' 0 4 63 250
P6-30 466033 1 91' 0 3 63 250
P6-60 466033 1 176' 0 3 63 250
P6-90 466033 1 261' 0 3 63 250
P6-120 466033 1 346' 0 3 63 250
#---- EXABYTE EXB-8200 MEDIA (P5)
P5-15 466033 1 69' 0 4 63 250
P5-30 466033 1 128' 0 3 63 250
P5-60 466033 1 246' 0 3 63 250
P5-90 466033 1 366' 0 3 63 250
#---- EXABYTE EXB-8500 MEDIA (P6)
X_256_H 932066 1 48' 0 7 63 500
X_512_H 932066 1 91' 0 5 63 500
X_1024_H 932066 1 176' 0 5 63 500
X_1536_H 932066 1 261' 0 5 63 500
X_2048_H 932066 1 346' 0 5 63 500
P6-15_H 932066 1 48' 0 7 63 500
P6-30_H 932066 1 91' 0 5 63 500
P6-60_H 932066 1 176' 0 5 63 500
P6-90_H 932066 1 261' 0 5 63 500
P6-120_H 932066 1 346' 0 5 63 500
#---- EXABYTE EXB-8500 MEDIA (P5)
P5-15_H 932066 1 69' 0 7 63 500
P5-30_H 932066 1 128' 0 5 63 500
P5-60_H 932066 1 246' 0 5 63 500
P5-90_H 932066 1 366' 0 5 63 500
From: mark@maui.Qualcomm.COM (Mark Erikson)
People keep asking this question. I also was concernec about it when
I first started doing this stuff. But, all you really need is a definition
which will allow 1 file system to not want more than one tape. You have
a tape length which is larger than most file systems. There are some
2GB disks out there and some stripped file systems which are larger than
2GB. I would say that you only need parameters as large as your largest
fs (imho).
Hi,
There have been a couple of sunmmaries on this list recently.
I'll attach them in case they are what you were after.
<attached uuencoded list deleted - Probably the same as Celeste's >
From: ept@eptsun1.ctd.ornl.gov (E P Tinnel)
EXB-8200 params we have been using for several years are:
/etc/dump 0ufbsd /dev/nrst1 100 6000 54000 /dev/rsd2a
I have an active query into Sun for the 'blessed' params
for the EXB-8500.
From: era@niwot.scd.ucar.EDU (Ed Arnold)
hennig@darmstadt.gmd.de recently summarized on the 8500, saying one
should use s=10380,d=4100000,b=112 on rst[8-15].
For the 8200, we use exactly the parms you gave above.
Comments
Got something to say?
You must be logged in to post a comment.
