Amanda-Users

Re: Running out of tape ...

2005-05-17 17:09:42
Subject: Re: Running out of tape ...
From: Vlad Popa <vlad.popa AT sbg.ac DOT at>
To: amanda-users AT amanda DOT org
Date: Tue, 17 May 2005 22:57:48 +0200
Hi Jon !

On Tue, May 17, 2005 at 07:17:08PM +0200, Vlad Popa wrote:
Hi Folks !
After spending some thrilling days  with amanda I got her running somehow..
Nice Lady, but see ..

I am using the vtape filedriver and my backupdisks are on a mounted ftp server using lufs. the shares of this server are mounted and owned by amanda. I have labeled 8 vtapes of 10Gb, my holding disk is set also to 10gb (holdingdisk 10240 megabytes ..). Dumpcycle is set to 7 days, tapecycle to 8 tapes , runspercycle to 7 bumpdays to 1. The ftp access is a bit slowly. When trying to write a file (with touch , ie as amanda ) on a share it takes 1 or 2 seconds untill it is finished or committed (the prompt is appearing again..)

The homes of the tapeserver are to be backed up, each home set in the disklist as an extra entry followed by "/etc" and /opt/htdocs (the apache files..) /opt/bdc (virus scanner)
here the disklist:

h50234 /etc          user-tar
h50234 /opt/htdocs   root-tar
h50234 /opt/bdc      root-tar
h50234 /home/albin   high-tar
h50234 /home/astrid  comp-root-tar
h50234 /home/aurel   root-tar
h50234 /home/diana   high-tar
h50234 /home/oeko    root-tar
h50234 /home/sawa    root-tar
h50234 /home/vlad    high-tar

The prior tests were made with the whole FQDN hostname incuding its domain part.
Backups are made daily at 3.00 am.

There were not many to no changes of size and files of the directories refering the DLE during the 1st backup untill today.

1) Although it is plenty of tape, holding disk, some parts are omitted to written to tape, not even taken at level one to tape . I get THIS complaint in the reports ..

FAILURE AND STRANGE DUMP SUMMARY:
h50234.ser /home/astrid lev 0 FAILED [out of tape]


I was previous amflush the holding disk home by hand to tape. The destination "tape" jumped after this action to the next number.

On the next day the tiny /etc was omitted  with the same failure report ..


Any ideas how to get to a fix ?

2) A question about the holding disk: amanda writes to it as a buffer, the data gets hopefully to tape, but will the disk be ever cleared of its contents and if so, when ?


Whenever taper successfully writes a dump from the holding disk to
"tape" it removes the original from the holding disk.


You show error messages with little or no context.  Were other
things succeeding?  Are these the only messages that show errors?

These were just the errors... The other entries made it on tape successfully. I did'n want to fill up the list with too much logs .. All of the entries in the DLE went to the holding disk and all but the "failure-notice" ones made it on tape. . The ftp share usage is 10 percent of the max. The holding disk is half full. But network could indeed be very slow and timeout. (touch of a empty file there needs 2 secs 'till the next propt arrives!) Sometimes, yes, it complaints about "short write" runnig out of tape What timeout would you purpose to set in amanda.conf for the tape access ?
Greetings

Vlad  Popa


<Prev in Thread] Current Thread [Next in Thread>