Summary : 5 gig Exabyte problems.

From: richard@hydres.uucp
Date: Tue Sep 15 1992 - 00:39:14 CDT


Thanks for all of your replies on how to write 5 gigs to a high density
exabyte. We now have a machine that works. There were dozens of replies
so to summarise I have just picked out the key points.

a) Always use new tapes, especially with earlier releases of SunOS, ie)
4.0.* or 4.1.1. If you use a tape thats been used on a low density drive,
you will not be able to write in high density mode to it.

b) If your 5 gig exabyte machine has been writing in 2.4 gig, low density mode,
you will need to power cycle it as SUn's driver is flakey about switching
density modes.

c) 2 gigs is the file size limit under SUnOS, so consequently that is the
maximun amount that a single file can be written to tape.

d) rst[0-7] are for low density and rst[8-15] are for high density.

e) The figures that we now use for a level 0 dump are :

   dump 0ubdsf 126 54000 13000 /dev/rst8

However everyone else seemed to have there own. I can't say if they work,
but I know ours do for us. Here are a few other alternatives :

   dump 0fubsd 126 108000 6000 /dev/nrst9

   dump 0ubdsf 126 54000 6000 /dev/rst8

   dump 0ubdsf 126 54000 6000 /dev/nrst0

   dump 0ubdsf 126 54000 12000 /dev/rst8

   dump 0bdfsu 126 54000 /dev/nrst8 226000

   dump 0dsbuf 54000 24000 126 /dev/nrst8

   dump 0bdfsu 126 54000 14500 /dev/nrst8

At the end of the day, it seems that you should be safe with

     b=126
     d=54000
     f= /dev/rst8
 
However there is some discrepancy with s the size parameter. This though is
only used to estimate the amount of tape required and to shout when it thinks
it has run out of tape. To calculate the size, the following equation has
been suggested.
 
size = ((512 * bf ) + 1920) * (ts - 2048) / (bf * 10667)
 
where :
 
bf = blocking factor (eg 126)
ts = 2.3*G or 5*G (less for shorter or partially filled tapes)
 
eg) for a 5 gig tape
 
size = (512*126 + 1920) * (5*1024*1024 - 2048) / (126 * 10667)
 
     = 259039

I'll leave you all to decide what you think is right. I am sticking
with 13000 cos it works.



This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:06:49 CDT