Thanks Gene and Alan,

Thank you for your recommendations.

Seems that increasing of holding disk space does the trick, though I didn't
want to go that way.

It's still not clear for me why "shoe shining" happens as the biggest
filesystem is located on the same server to which the tape drive is
connected to.
In my understanding a data stream should be passed to the taper using
loopback interface so it should have enough throughput for direct backup to
the tape.

Kind Regards,
Dmitry

-----Original Message-----
From: [email protected] [mailto:[email protected]]
On Behalf Of Gene Heskett
Sent: Tuesday, July 14, 2015 4:36 PM
To: [email protected]
Subject: Re: amanda 3.3.7 intermittent issue



On Monday 13 July 2015 22:13:44 Alan Hodgson wrote:
> On Tuesday, July 14, 2015 01:35:33 PM you wrote:
> > Quite often backup fails with the errors:
> >
> >
> >
> > FAILURE DUMP SUMMARY:
> >
> >   backup-server /backup lev 0  FAILED [data write: Broken pipe]
> >
> >   backup-server /backup lev 0  partial taper: No space left on 
> > device, splitting not enabled
> >
> >   backup-server /backup lev 0  FAILED [data write: Broken pipe]
> >
> >   backup-server /backup lev 0  FAILED 1 tapes filled; runtapes=1 
> > does not allow additional tapes
> >
> >
> >
> > It looks strange as there should be plenty space left on the tape
> > (LTO6 - 2.5TB for uncompressed and 6.25 TB for compressed).
>
> In my experience, if tapes are filling up too early, it's likely a 
> sign that the tape drive has a hardware problem (assuming the tapes 
> are good and you're cleaning the drive regularly).
>
> You could experiment with dd'ing large files to it and you'll probably 
> see it give end-of-tape at seemingly random amounts of data.
>
> > Other issue is that this backup-server:/backup filesystem is the 
> > biggest one and it is backed up directly to the tape without using 
> > holding disk.
> >
> > The issues here is that despite the settings that I have in
> > amanda.conf:
> >
> > dumporder "ssssssssss"
> > taperalgo first
>
> I suspect that not using a holding disk is causing these settings to 
> not work as expected, but hopefully someone who knows for sure will 
> chip in.

If not using a holding disk causes the tape drive to shoe shine, its been my
experience that all bets are off. 

So I arranged to have a big enough holding disk area on the main drive by
moving it to the largest partition, and switched to vtapes on a separate
hard drive, but did not do this at the same time.  Its been very dependable
since as long as I don't do any swap recovery.  My system has used up
several HD's since, but smartctl gives me plenty of warning about impending
drive problems, so the failing drive can always be replaced long before
amanda gets a tummy ache.

I most certainly cannot say that for the tape drives I attempted to use
prior to going vtapes.

A side effect of vtapes, is that recovery/restore operations are hundreds of
times faster since the vtape is not sequential, but is truly random access.
I consider that a huge advantage.  Not having the ability to easily do
offsite backups seems a valid tradeoff.  I could, as those drives are in a
hot swap cage, but just never got a round tuit.

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>


-- 
Disclaimer:  Asparona is part of the Deloitte group of companies. Deloitte
refers to the New Zealand member firm of Deloitte Touche Tohmatsu Limited.

The information contained in this document is confidential to the
addressee(s) and may be legally privileged. Any view or opinions expressed
are those of the author and may not be those of the organisation to which
the author belongs. No guarantee or representation is made that this
communication is free of errors, viruses or interference. If you have
received this e-mail message in error please delete it and notify me. 

Reply via email to