[nycbug-talk] physst

Kurt Starsinic kstarsinic
Wed Aug 11 15:06:45 EDT 2004


On Wed, 11 Aug 2004 14:31:36 -0400, Joshua S. Freeman <jfreeman at nybg.org> wrote:
> interesting..
> 
> well, what's supposed to be happening is that the taptest utility associated
> with bacula should be doing a 'rawfill' of data onto a tape in a
> tapedrive...

    It's only a problem if it *freezes* in that state.  For something
that's doing relatively raw I/O to a relatively slow device, I would
*expect* the process to be mostly "waiting for the hardware" until
taptest completes.

    When I said that it "usually becomes *of interest*" with a process
that can't get killed, I meant it's one of the very few ways that a
process can (theoretically, and even then only in the case of device
driver bugs) get into an unkillable state.

    - Kurt




More information about the talk mailing list