On Tue, Aug 07, 2001 at 04:12:49PM -0400, Larry Kingery wrote:
> NetBackup will write until it receives an end-of-tape message, then
> load up another tape and continue. A single backup does not need to
> fit on a single tape.
OK, now it's clear, thanks.
>
> However, there is a bp.conf entry you can add called
> DISALLOW_BACKUPS_SPANNING_MEDIA. I'm not sure what error you'd get
> with this, I've never had reason to use it.
>
I don't use this setting.
> A DLT8000 drive with DLT IV tapes (as printed on the cartridge itself,
> not to be confused with Media Manager's dlt, dlt2, and dlt3 tape
> "types") will hold 40GB of uncompressed data. There's really no way
> to say how much it will hold when using compression. You can get an
> estimate by using the available_media script and averaging the usage
> of all the FULL tapes.
>
> The tape selection algorithm is described beginning on page 638 of the
> NBU System Administrator's Guide (UNIX 3.4).
>
> If you find this happening on a single tape, I would recommend
> freezing the tape so that it is not used again:
>
> bpmedia -freeze -ev <mediaid>
That was my first reaction. I thought it was single tape, so I froze it.
Unfortunately, it's the pattern on all tapes.
>
> If this happens with most/all tapes, the first thing I would look at
> is to make sure the system is configured correctly for that drive
> type. Reference the Media Manager Device Configuration Guide.
OK, I am not an expert in this area.
So, the drive is Sun StorEdge L9. Media Manager configured to use drive type
DLT2
for it. Is it correct setting?
And this is the error pattern:
03:07:10 [12727] <2> signal_parent: sending SIGUSR1 to bpbrm (pid = 12725)
03:07:10 [12727] <2> write_data: attempting write error recovery, err = 5
03:07:10 [12727] <2> tape_error_rec: error recovery to block 1221860 requested
03:07:10 [12727] <2> tape_error_rec: attempting error recovery, delay 3 minutes
before nex
t attempt, tries left = 5
03:10:10 [12727] <2> io_ioctl: command (0)MTWEOF 0 from (overwrite.c.387) on
drive index 0
03:10:10 [12727] <2> io_ioctl: MTWEOF failed during error recovery, I/O error
03:10:10 [12727] <2> tape_error_rec: cannot read position for error recovery,
read_positio
n scsi command failed, key = 0xb, asc = 0x44, ascq = 0x89
03:10:10 [12727] <2> tape_error_rec: attempting error recovery, delay 3 minutes
before nex
t attempt, tries left = 4
The next attempt:
03:40:13 [12832] <2> signal_parent: sending SIGUSR1 to bpbrm (pid = 12830)
03:40:13 [12832] <2> write_data: attempting write error recovery, err = 5
03:40:13 [12832] <2> tape_error_rec: error recovery to block 1206730 requested
03:40:13 [12832] <2> tape_error_rec: attempting error recovery, delay 3 minutes
before nex
t attempt, tries left = 5
03:41:05 [12940] <2> bptm: INITIATING: -count -cmd -rt 8 -rn 0 -stunit
L9_Library -den 15
-mt 2 -masterversion 340000
03:41:05 [12940] <2> bptm: EXITING with status 0 <----------
03:41:06 [12942] <2> bptm: INITIATING: -count -cmd -rt 8 -rn 0 -stunit
netbackup-dlt2-robo
t-tld-0 -den 15 -mt 2 -masterversion 340000
03:41:06 [12942] <2> bptm: EXITING with status 0 <----------
03:43:13 [12832] <2> io_ioctl: command (0)MTWEOF 0 from (overwrite.c.387) on
drive index 0
03:43:13 [12832] <2> io_ioctl: MTWEOF failed during error recovery, I/O error
03:43:13 [12832] <2> tape_error_rec: cannot read position for error recovery,
read_positio
n scsi command failed, key = 0xb, asc = 0x44, ascq = 0x89
03:43:13 [12832] <2> tape_error_rec: attempting error recovery, delay 3 minutes
before nex
t attempt, tries left = 4
I got a suspicion that end of tape is not detected properly. Is there any way I
can test it?
Also, could it be defective media? (All cartridges defective?)
>
>
> Konstantin 'Kastus' Shchuka writes:
> > Hi NetBackup gurus,
> >
> > I have a stupid problem due to the lack of my understanding how NetBackup
> > works.
> >
> > I have a Sun StorEdge L9 tape autoloader. It writes to DLT2 tapes.
> > I have a class for full back up of one server which takes appr. 30G.
> > One tape can store 40G, if I understand it correctly.
> >
> > THe first backup goes to tape without problems. When the second backup is
> > attempted,
> > the tape writes to some point, then returns write media error, probably
> > because of the end of the tape reached. So the attempt fails with code 84.
> >
> > Then another attempt starts, but the same tape is being selected and the
> > story
> > repeats.
> >
> > My question is: should one backup fit on one tape? Or can it start on one
> > tape
> > and continue on another? How does NetBackup estimate what can go to some
> > particular tape? Is there any way to prevent such behaviour?
> >
> > Thank you,
> > --
> > Konstantin 'Kastus' Shchuka
> > Unix System Administrator
> > ePocrates Inc.
> > tel 650.232.4886
> > fax 650.592.6995
> > _______________________________________________
> > Veritas-bu maillist - Veritas-bu AT mailman.eng.auburn DOT edu
> > http://mailman.eng.auburn.edu/mailman/listinfo/veritas-bu
>
> --
> Larry Kingery
> 3 kinds of people: those who can count & those who can't.
--
Konstantin 'Kastus' Shchuka
Unix System Administrator
ePocrates Inc.
tel 650.232.4886
fax 650.592.6995
|
