Re: [Dump-users] Solved: Problems spanning tapes with dump (summary)

[Mike C. <da...@ix...>]

On Fri, Nov 21, 2003 at 10:29:33AM -0800, Tom Haws wrote:
> detected by dump, it calls the script, which does an rsh back to the 
> host with the tapedrive, rewinds the tape there (causing our autochanger 
> in sequential mode to load the next tape), sleeps a couple of minutes, 
> then exits with a return code of zero.  This allows dump to resume, 
> without trying to contact anyone, and all is fine.


A 'sleep 2 minutes' approach always scares me.  No matter how long you make
the delay, sometime there is going to be a case where it's not long enough
and things will crash.

Would a better solution be to make use of mt status on the drive
periodically and when it's done changing the tape, exit?   By making the
call to mt status something like every 10 seconds, you might actually
return in less than 2 minutes in most cases, and increase throughput!  :->

Obviously you'd want to put limits on how many times you run mt status so
you can detect that things went really bad.

mrc
-- 
     Mike Castle      da...@ix...      www.netcom.com/~dalgoda/
    We are all of us living in the shadow of Manhattan.  -- Watchmen
fatal ("You are in a maze of twisty compiler features, all different"); -- gcc