On Friday 25 July 2014 18:00:35 Gene Heskett did opine
And Gene did reply:
> On Friday 25 July 2014 16:49:02 Debra S Baddorf did opine
> 
> And Gene did reply:
> > On Jul 25, 2014, at 3:15 PM, Gene Heskett <[email protected]> wrote:
> > > On Friday 25 July 2014 15:26:00 Debra S Baddorf did opine
> > > 
> > > And Gene did reply:
> > >> I just create two DLEs
> > >> 
> > >> mynode  /         backuptype
> > >> mynode  /boot  backuptype
> > >> 
> > >> Then   amdump  myconfig           (  or  amdump  myconfig  mynode 
> > >> ) will get both of them.    Is this not an option for you?
> > > 
> > > That would completely hose amanda's ability to spread the levels of
> > > the backups out over the cycle in its attempt to make full use of
> > > the backup media available.
> > 
> > Huh?  As I see it,  it *allows*  amanda to spread the levels around
> > to different days.    BUT it is true that “mynode”  may not have
> > both pieces done at level 0 on the same day.
> > 
> > > I could do it, yes, but that is not what amanda is all about.
> > 
> > I thought it was EXACTLY what amanda was all about.
> > 
> > > It would not be a huge problem with vtapes, but for those using
> > > individual tapes, that would never fly unless you allowed amanda to
> > > use a virtually unlimited number of tapes when it is doing a level
> > > 0 backup.
> > 
> > I do use real tapes, not vtapes.   But “doing a level 0 backup” as
> > you put it implies you are trying to do them all on one day - which
> > isn’t what Amanda is about.   I have 30 or so nodes, each with 3-8
> > DLEs. The level 0s get spread all around the week  (doing backups
> > each day, with level-0 once per week at minimum).
> 
> If by "nodes" you are referring to the individual machines in your
> responsibility tree, then yes, we agree. With only 3 such nodes to pick
> from, and with the widely varying amounts of data, its not quite as
> practical.  With only 6 DLE's, I'd have 5 backups of around a gigabyte
> maximum, and one at around 350Gb.
> 
> But Jon tells me the softlinks are backed up, while my amrecover of one
> subdir with all the important gcode and machine configuration files in
> it, less than a week ago, restored exactly zero of the 3 or 4
> softlinks in that lathe:/home/gene/linuxcnc subdir.  I had to
> re-create them from scratch.  If they were in the backup, and the log
> (as I posted earlier says they weren't, they were skipped,) then
> amrecover should have re- created them for me.  It did not because
> they were not in the backup, they were skipped due to the perceived by
> tar, different filesystem error. Thats the error I'm preaching about,
> that link did NOT point to a different filesystem, but tar says it
> did.
> 
> And that is the bug I am standing on the stump yelling about.
> 
FWIW, I took the --one-filesystem option out of the tar invocation last 
night, which got rid of the warnings, but then had a glitch in the 
estimate phase:

Got empty header from coyote/usr/share,

but the "by dle" report says it did a level 0 just fine.  Me wanders off 
scratching head. I'll see if its repeated tonight.

Cheers, Gene Heskett
-- 
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
Genes Web page <http://geneslinuxbox.net:6309/gene>
US V Castleman, SCOTUS, Mar 2014 is grounds for Impeaching SCOTUS

Reply via email to