Re: DLE splitting

[Gene Heskett]

On Saturday 18 September 2021 16:21:39 Jon LaBadie wrote:

> On Fri, Sep 17, 2021 at 11:55:09AM -0400, Jon LaBadie wrote:
> >I'm seeing instances of wasted VTape with amanda switching
> >to a new tape for the last DLE even though there is room
> >left on the current tape.
> >
> >Here is the relevant section from yesterday.
> >
> >USAGE BY TAPE:
> >  Label Time Size  %  DLEs Parts
> >  DS1-010   0:08   10884M   10.63131
> >  DS1-011   0:51   74205M   72.5 115
> >
> >If combined on one tape it would have only filled 83% of the
> >tape.  Earlier this month the numberw were worse, 9% and 66%.
> >Even if a second tape were needed, the last DLE should have
> >at least started on the first tape.
>
> A new thought.  When the amanda taper has been writing to a
> vtape and attempts to open the next part file, is there some
> timeout that would look like the vtape is full/unavailable?
>
> Four of my six vtape drives, including the one in use above,
> are in an external USB enclosure.  These drives go to sleep
> if inactive for a period.  When a drive is accessed, it takes
> 20 to 30 seconds for the 4 drives to spin up and be ready.
>
> The DLE that was not taped on the first vtape took over an
> hour to dump.  Certainly the USB box went to idle state
> before it was ready to tape.
>
> When I do things from the command line that awakens the idle
> USB box, the command just blocks until the device is active
> (as I say, 20-30sec) and then completes normally.
>
> But perhaps the calls (perl or C) that amanda uses to open
> the next part return with a code that says EOF or error
> rather than "currently unavailable, try again".
>
> Jon

A sensible suspicion. But I am w/o facts to test.

Let me tell a story. One I should have told months ago because I sort of 
left Nathan hanging.  Bad form. 

At the same time I replaced the 1T /amandatapes drive with a 2T, both 
spinning rust so the same speed, I moved the holding disk from spinning 
rust to a small SSD, which is about 5x faster than rust. All my crc 
errors vanished too.

My backups used to take 2.5 to 3.5 hours, are now done in around half an 
hour thanks to the faster holding disk. Last nights was done in 22 
minutes.

Copyright 2021 by Maurice E. Heskett
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)
If we desire respect for the law, we must first make the law respectable.
 - Louis D. Brandeis
Genes Web page 

Re: DLE splitting

[Jon LaBadie]

On Fri, Sep 17, 2021 at 11:55:09AM -0400, Jon LaBadie wrote:

I'm seeing instances of wasted VTape with amanda switching
to a new tape for the last DLE even though there is room
left on the current tape.

Here is the relevant section from yesterday.

USAGE BY TAPE:
 Label Time Size  %  DLEs Parts
 DS1-010   0:08   10884M   10.63131
 DS1-011   0:51   74205M   72.5 115

If combined on one tape it would have only filled 83% of the
tape.  Earlier this month the numberw were worse, 9% and 66%.
Even if a second tape were needed, the last DLE should have
at least started on the first tape.



A new thought.  When the amanda taper has been writing to a
vtape and attempts to open the next part file, is there some
timeout that would look like the vtape is full/unavailable?

Four of my six vtape drives, including the one in use above,
are in an external USB enclosure.  These drives go to sleep
if inactive for a period.  When a drive is accessed, it takes
20 to 30 seconds for the 4 drives to spin up and be ready.

The DLE that was not taped on the first vtape took over an
hour to dump.  Certainly the USB box went to idle state
before it was ready to tape.

When I do things from the command line that awakens the idle
USB box, the command just blocks until the device is active
(as I say, 20-30sec) and then completes normally.

But perhaps the calls (perl or C) that amanda uses to open
the next part return with a code that says EOF or error
rather than "currently unavailable, try again".

Jon

--
Jon H. LaBadie j...@jgcomp.com
 11226 South Shore Rd.  (703) 787-0688 (H)
 Reston, VA  20190  (703) 935-6720 (C)

Re: allow-split configuration parameter

[Gene Heskett]

On Saturday 18 September 2021 02:26:23 Jon LaBadie wrote:

> On Fri, Sep 17, 2021 at 01:09:48PM -0400, Gene Heskett wrote:
> >On Friday 17 September 2021 12:09:55 Jon LaBadie wrote:
> >> While investigating a tape issue (separate thread)
> >> I investigated the "allow-split" parameter of
> >> amanda.conf.  The man page says it defaults to "true"
> >> and my configuration has never set it.
> >>
> >> Thinking perhaps the default has changed I tried
> >> setting allow-split several ways and places.  amcheck
> >> always complains "configuration parameter expected".
> >>
> >> Also, amgetconf says there is no configuration parameter
> >> called "allow-split".
> >>
> >> Does anyone have "allow-split" in their configuration?
> >> If so, how/where?
> >
> >I do acc grep:
> >gene@coyote:~$ sudo grep allow
> > /usr/local/etc/amanda/Daily/amanda.conf maxdumps 1  # allowing
> > one per host makes sense to me
> >#   skip-full - skip the disk when a level 0 is due, to allow full
> >backups
> >
> >define dumptype  global {
> >...
> > allow-split true
> >}
> >
> >
> >Copyright 2019 by Maurice E. Heskett
> >Cheers, Gene Heskett
>
> OF COURSE, it a dumptype parameter.   My stupidity.
>
> I've added it, but I suspect it will have no affect
> as the default is true.
>
> Jon
Thats my assumption too Jon. But when I put the 2T drive in 
as /amandatapes, I expanded the size of the vtapes, and its not skipped 
to the next tape since.  That was 7 or 8 tapecycles back up the log. The 
lady is almost civilized now. :)

I just added the /home tree of another machine as I now have one 
delegated to running cura and openscad so I've a one stop src to drive a 
3d printer which is making me a magazine stuffer so that 6.5 Creedmoor 
shells might feed thru a P-17 action which is what old AT is built on, 
but after wearing out 4 barrels chambered for the Ackley-improved 30-06 
since I built it in the '60's, I screwed a 6.5 barrel in it 2 years back 
and chambered it for the Creedmoor, a bit easier on the old mans 
shoulder since it will be 87 in a couple weeks.  But the shorter, fatter 
shell made a single shot out of it, so I'm trying to fix it...

Since that machine isn't surrounded by human comfort first furniture, I 
log into it from here to program OpenSCAD via an ssh -Y login. Its 
running debian buster and can run the latest AppImage of OpenSCAD 
whereas this machine is still running stretch and won't run the latest 
OpenSCAD. But I am preparing to put bullseye on this one, and have a 
large pocketfull of 1T SamSung 870 SSD's I'm going to use as a raid 6 or 
raid 10, for everything but /boot & swap with a separate 1/2T 860 
SamSung drive for /boot & swap.

So in another 2 or 3 weeks I should have bullseye up and running here, so 
the folks on the debian list will quit giving me hell for running old 
stuff. But with the OpenSCAD exception its running very well.

I hope this finding you and Gundy in good health.

My Dee passed from COPD last Pearl Harbor Day, so I'm alone now. And the 
place is keeping me busy. Replaceing split rail fencing that has to be 
rebuilt every 10 years with vinyl, that sort of stuff. The shasta white 
shingles I put on in 2010 after the direcho remodeled the place have 
lost the white, running the AC bill way up, so thats gotta be redone 
next spring.  We'll see how I feel then.

Take care you two.

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)
If we desire respect for the law, we must first make the law respectable.
 - Louis D. Brandeis
Genes Web page 

Re: allow-split configuration parameter

[Jon LaBadie]

On Fri, Sep 17, 2021 at 01:09:48PM -0400, Gene Heskett wrote:

On Friday 17 September 2021 12:09:55 Jon LaBadie wrote:


While investigating a tape issue (separate thread)
I investigated the "allow-split" parameter of
amanda.conf.  The man page says it defaults to "true"
and my configuration has never set it.

Thinking perhaps the default has changed I tried
setting allow-split several ways and places.  amcheck
always complains "configuration parameter expected".

Also, amgetconf says there is no configuration parameter
called "allow-split".

Does anyone have "allow-split" in their configuration?
If so, how/where?


I do acc grep:
gene@coyote:~$ sudo grep allow /usr/local/etc/amanda/Daily/amanda.conf
maxdumps 1  # allowing one per host makes sense to me
#   skip-full - skip the disk when a level 0 is due, to allow full
backups

define dumptype  global {
...
allow-split true
}


Copyright 2019 by Maurice E. Heskett
Cheers, Gene Heskett


OF COURSE, it a dumptype parameter.   My stupidity.

I've added it, but I suspect it will have no affect
as the default is true.

Jon

--
Jon H. LaBadie j...@labadie.us
 11226 South Shore Rd.  (703) 787-0688 (H)
 Reston, VA  20190  (703) 935-6720 (C)