Subject: Re: restore(8)/tape quirk in 1.4?
To: Gandhi woulda smacked you <greywolf@starwolf.com>
From: Brett Lymn <blymn@baea.com.au>
List: current-users
Date: 05/19/1999 20:28:23
According to Gandhi woulda smacked you:
>
>On Mon, 17 May 1999, Hauke Fath wrote:
>
># 
># # What is this "user buffer"?
># 
># # Why is 1.4 restore(8) incapable of dynamically adjusting to a tape's
># (fixed) buffer size -- contrary to what the man page says? 

Because it's _always_ been incapable of doing it.  The man page is
wrong, NetBSD is not the only OS that has this problem SunOS had it -
I have not checked Solaris but I would almost bet money it too is
broken.  Either restore or the man page should be fixed :-)

>#Why is it
># incompatible to pre-1.3 NetBSD dumps - without comment?
>
>Yow.  Haven't hit this one yet.
>

Neither have I.  I will check this one....

>
>On another tangent, on a disk-to-disk dump, I get 370KB/sec.  I thought
>transfer rates were supposed to be in the megabyte range?
>

Probably because of the synchronous meta-data writes.  Have you tried
restoring to a file system mounted -async and see what throughput you get?

-- 
===============================================================================
Brett Lymn, Computer Systems Administrator, British Aerospace Australia
===============================================================================