On Fri 06 Sep 02 16:59, Jon LaBadie wrote:
> On Fri, Sep 06, 2002 at 11:01:45AM +0200, Brian Jonnes wrote:
> > Pulled the tape out, plugged it back in, and now its dumping fine. Can
> > anyone make sense of this?
>
> I don't use your systems but ISTR people mentioning using mt to set the
> block size to 32K.

Tried that just now; failed.

Now the strange thing is that I'm getting _partial_ success now. The "taper" 
line in NOTES gives me (as an example):

  taper: tape DAILY_02 kb 467776 fm 48 writing filemark: Input/output error

Where previously it was giving me kb 0 (with ide-tape). Again I stress, some 
of the dumps finish fine. I got just under 2GB written on Thursday morning, 
but Friday and Saturday have now failed (Friday's got 64kb written; tried an 
amflush and it got 35MB written).

I'm now guessing that ide-tape was merely _reporting_ that 0k was written; 
ide-scsi is now being more accurate with regards to how much data is written 
before failure.

I think I'm gonna throw this drive against the wall! As I've mentioned, I 
already had the unit swapped out. Right now I can't imagine it being anything 
else, 'cause the errors are distributed throughout the tape cycle.

..Brian

> > BTW is there a way of getting amanda to put level 0's in "degraded mode"?
> > I'd still like full dumps when I have to manually run amdump.
>
> level 0 and degraded mode?  Aren't these non-sequitors?  I thought degraded
> mode meant falling back to incrementals because there was no place to put
> the level 0's.  I.e. the tape was not working and the holding disk did not
> have enough unreserved space to hold the level 0's.  So where would you
> like amanda to put your huge level 0's if it can't put it to tape or to
> disk?
>
> What's your holding disk space and reserve % situation?

-- 
Init Systems  -  Linux consulting
031 767-0139    082 769-2320    brian AT init.co DOT za