I've seen this problem before. Are your tape drives shared by multiple
media servers and if so, do you have SSO turned on? It looks like the
tape was possibly overwritten and no longer has a valid volume header.
Here's what can happen:
In a shared tape environment, when one media server is using a tape
drive, NBU (with SSO) will issue a SCSI reserve command to the drive so
that it has exclusive use of the drive. If it does not do this, other
media servers can interfere with the tape drive, possibly rewinding the
tape to the beginning while the media server doing the backup is writing
to the tape. The media server doing the backup is unaware that anything
has happened and happily overwrites data it has already written.
Unfortunately, there were bugs in the implementation, and the SCSI
reserve wasn't happening. If you're using NBU 3.4, make sure you have
installed patch 5 as the SCSI reserve bug still exists for HPUX in patch
4. Not sure about NBU 4.5. Also, make sure that the file
/usr/openv/netbackup/db/config/ENABLE_SCSI_RESERVE exists. It's just an
empty file.
Of course, if you're not in a shared tape drive environment, the SSO
stuff doesn't apply to you, but you may have had another application
attempting to use the tape drive at the same time as NBU.
On Wed, 2003-09-03 at 06:41, dwlee wrote:
> Hello,
>
> Environment
>
> ===============================================================================================
>
> Master Server : HP-UX 11i
>
> Media Server : HP-UX 11i
>
> Tape Library : STK L700 with 9840 drive, 5 Drivers
>
>
>
>
>
> Symptom
> ===================================================================================
>
> When we tried to restore from tape, it happed failed.
>
> Then we saw bptm log of media server, it happed errors as below.
>
>
>
> 13:41:56.514 [16110] <2> getsockconnected: host=nbu_master
> service=bpdbm address=47.96.8.13 protocol=tcp non-reserved port=13721
>
> 13:41:56.515 [16110] <2> logconnections: BPDBM CONNECT FROM
> 47.96.7.66.64296 TO 47.96.8.13.13721
>
> 13:41:56.796 [16110] <16> io_read_media_header: block read is not a
> NetBackup media header, len = 32768, media id SC0159, drive index 2,
> data is unknown
>
> 13:41:56.796 [16110] <2> check_error_history: called from bptm line
> 10235, EXIT_Status = 172
>
> 13:41:56.796 [16110] <2> io_close: closing
> /usr/openv/netbackup/db/media/tpreq/SC0159, from bptm.c.12946
>
>
>
> What do you mean above errors?
>
> When do you happed in the situation or environment ?
>
> And What do I action how to solve this problem ?
>
>
>
> Thanks.
>
>
--
Jack L. Forester, Jr.
Sr. UNIX Systems Administrator
Lockheed Martin Information Technology
(304) 625-3946
|
