xfs
[Top] [All Lists]

xfsdump/xfsrestore problem - unable to restore entire backup

To: linux-xfs@xxxxxxxxxxx
Subject: xfsdump/xfsrestore problem - unable to restore entire backup
From: jansen <jansen@xxxxxxxxxxxxxx>
Date: Fri, 25 Jan 2002 09:38:59 -0600
Sender: owner-linux-xfs@xxxxxxxxxxx
User-agent: Mozilla/5.0 (X11; U; Linux i586; en-US; rv:0.9.7) Gecko/20011226

Hi,

I recently needed to restore some data from an old backup and ran into two
problems. One problem was annoying but no big deal, the other is a real
problem.

First problem:

I started an interactive restore from a linux machine using a remote
tape drive hooked to an O2 using the session id, xfsrestore found
the dump and gave the normal interactive command line.  I found the
file I wanted to restore and "add"ed it and then did "extract".  xfsrestore
then immediately returned with a status of SUCCESS after only having
created the directories and not restoring the file I had selected.

Second problem:

I started a non-interactive xfsrestore session in the same way I
did the above and it once again found the xfsdump session and it
extracted a couple gigs of data before it gave this error:

xfsrestore: restoring non-directory files
xfsrestore: examining media file 1
xfsrestore: seeking past media file directory dump
xfsrestore: drive_scsitape.c:1463: do_next_mark: Assertion `rechdrp->first_mark_
offset - rechdrp->file_offset <= ( off64_t )( contextp->dc_recsz )' failed.
Aborted (core dumped)

I also tried restoring this dump session directly from the O2 that
has the tape drive and it also failed with a SEGV but without the
assert error message (sorry I don't have the exact output).  I had
the same problem with an earlier backup of the same machine.

I did the non-interactive restore above again with
"-v drive=debug,media=debug" added to the command line and here is
the last 30 or so lines of the debug output:

xfsrestore: drive op: return read buf: sz 73728 (0x12000)
xfsrestore: drive op: read: wanted 32 (0x20)
xfsrestore: drive op: return read buf: sz 32 (0x20)
xfsrestore: drive op: get mark
xfsrestore: drive op: read: wanted 256 (0x100)
xfsrestore: drive op: return read buf: sz 256 (0x100)
xfsrestore: Media_end: pos==3
xfsrestore: drive op: end read
xfsrestore: tape op: get status
xfsrestore: tape status = fmk,wprot,onl
xfsrestore: encountered EOF attempting to read record 8849
xfsrestore: Media_mfile_next: purp==2 pos==0
xfsrestore: drive op: begin read
xfsrestore: tape op: get status
xfsrestore: tape status = fmk,wprot,onl
xfsrestore: tape op: reading 245760 bytes
xfsrestore: validating media file header
xfsrestore: media file header valid: media file ix 1
xfsrestore: examining media file 1
xfsrestore: file 1 in object 1 of stream 0
xfsrestore: file 10 in stream, file 1 of dump 0 on object
xfsrestore: dump session label: "joe"
xfsrestore: dump session id: a28f36fb-8d12-44a9-8df1-657ca2dc62af
xfsrestore: stream 0, object 1, file 1
xfsrestore: drive op: read: wanted 131072 (0x20000)
xfsrestore: tape op: reading 245760 bytes
xfsrestore: tape op: reading 245760 bytes
xfsrestore: drive op: return read buf: sz 4736 (0x1280)
xfsrestore: drive op: read: wanted 126336 (0x1ed80)
xfsrestore: tape op: get status
xfsrestore: tape status = fmk,wprot,onl
xfsrestore: encountered EOF attempting to read record 2
xfsrestore: using on-media session inventory
xfsrestore: drive op: end read
xfsrestore: drive op: get device class
xfsrestore: drive op: eject media
xfsrestore: tape op: closing drive


The Linux machine is currently running the stock XFS 1.0.2 install with
the (non-kernel) RedHat updates.  The Linux machine was most likely
running the XFS 1.0.1 version when it was backed up. The O2 is running
IRIX 6.5.14m.  The Linux machine is a VA Linux with (if I recall correctly)
a TYAN Thunder 2500 motherboard, 3GB of SDRAM and Seagate disks on the
integrated Symbios SCSI controllers.  The tape drive is a four tape
Seagate AIT jukebox.

The backup that created this dump did not have any error messages but
was split between two tapes.

The command line I used to backup the drive is:

/usr/sbin/xfsdump  -l 0 -T -d 2048 -o -L joe -Y 7 -f 
guest@O2:/dev/rmt/tps2d6nrvc /dev/sdc1

The command line I used to do the interactive restore is:

xfsrestore -i -S 87020d4e-4b28-4ddb-ab0a-54bd071d50c3 -f 
guest@O2:/dev/rmt/tps2d6nrvc .

I have the xfsrestore core file if it is useful.  I can supply any other
information needed.

Thanks for any assistance.


--


------- Stephan


<Prev in Thread] Current Thread [Next in Thread>
  • xfsdump/xfsrestore problem - unable to restore entire backup, jansen <=