Koozali.org: home of the SME Server
Legacy Forums => Experienced User Forum => Topic started by: Guck Puppy on July 23, 2003, 07:40:13 AM
-
I know, this topic in one form or another litters these foraa.
I read things like "change the drive", "change the tapes". Well this is on a server that was recently changed over from redhat 7.1 - with a working flexbackup system.
The server is not even getting close to performing a backup :
Jul 22 02:00:00 my-server e-smith[10826]: Processing event: pre-backup
[pre-backup stuff etc.]
Jul 22 02:00:01 my-server kernel: st0: Error with sense data: Current st09:00: sense key Illegal Request
Jul 22 02:00:01 my-server kernel: Additional sense indicates Invalid command operation code
Jul 22 02:00:01 my-server kernel: st0: Error with sense data: Current st09:00: sense key Illegal Request
Jul 22 02:00:01 my-server kernel: Additional sense indicates Invalid command operation code
Jul 22 02:00:01 my-server e-smith[10856]: Processing event: post-backup
[post-backup stuff etc.]
I have installed the "flexbackup" panel contrib, I changed it to use "afio" as the format though I suspect that is irrelevant since it's not even getting to the point of backing up. I have left the buffer at 10.
I've also installed the updated flexbackup rpm from Charlie Brady, taking it to version 1.1.6-1 and the mbuffer rpm - though I am not sure if flexbackup will 'just use it' by default. The "buffer" version remains at 1.19-5.
Anyone any ideas how I can get it to progress at least to giving me backup errors rather than "I can't even start to backup!" errors?
G
-
I think everyone must be trying to reply to this thread all at once and the phorum system is jamming :)
Seriously though, I know this topic is annoying - but if anyone who's using a Travan drive with 5.6 with Complete success could share their experiences, it would be very uplifting. OK, maybe not *complete* success, but anyone doing ok with it?
G
-
Hi GP,
If you login as root and run flexbackup directly, ie :-
flexbackup -fs all
at the console prompt, what do you get ?
Kelvin
-
Kelvin wrote:
> If you login as root and run flexbackup directly, ie :-
>
> flexbackup -fs all
>
> at the console prompt, what do you get ?
I am using the updated flexbackup v.1.1.6-1, so I get :
[root@my-server root]$ /usr/bin/flexbackup -fs all
Unknown option: fs
but when I try 'flexbackup -set all' I get :
[root@my-server root]$ /usr/bin/flexbackup -set all
flexbackup version 1.1.6 (http://flexbackup.sourceforge.net)
/etc/flexbackup.conf syntax OK
Errors:
$erase_tape_set_level_zero not defined
That doesn't sound very good... when browsing through the /etc/flexbackup.conf, I noted that it's set for the 'old style' -fs so I changed $erase_tape_all_level_zero to $erase_tape_set_level_zero, and added an $afio_nocompress_types like it asked and it did finally attempt to run the backup. (Oh, I had to install afio from an rpm as well. I like afio, I used it on the old 7.1 server setup).
However, the backup failed : (repetitive-esque parts are ... removed)
| printf "Volume Label:\nlevel 0 / Wed Jul 23 15:26:17 2003 afio+gzip from
| my-server\n\n" > /tmp/label.981
| cd "/" && (printf "//--/tmp/label.981 flexbackup.volume_header_info\n" &&
| find . -xdev -depth ! -regex "news/articles" ! -regex ".*~"$ -print ) |
| afio -o -E /tmp/nocompress.981 -z -P gzip -Q -4 -Z -M 2m -T 3k -v -a -b
| 10k - | mbuffer -m 3M -s 10k | dd ibs=10k obs=10k conv=noerror,sync
| of="/dev/nst0"
| rm -f /tmp/nocompress.981 /tmp/label.981
|------------------------------------------------------------
CONTROL_FILE/flexbackup.volume_header_info -- okay
lost+found -- okay
boot -- okay
dev/pts -- okay
dev/cdrom -- okay
dev/log -- okay
dev/null -- okay
dev/md10 -- okay
dev/md0 -- okay
dd: opening /dev/nst0': Read-only file system
0.0 kB/s in - 0.0 kB/s out - 0 kB total - buffer 0% fulldev/MAKEDEV.z -- (44%)
dev/md1 -- okay
...
dev/md9 -- okay
dev/mdspstat -- okay
...
dev/ataraid/d0p1 -- okay
...
dev/ataraid/d6p2 -- okay
dev/ataraid/d6p3 -- okay
dev/ataraid/d6p4 -- okay
dev/ataraid/d6p5 -- okay
afio: "-": Broken pipe
ERROR from backup, exiting
offending command(s):
cd "/" && (printf "//--/tmp/label.981 flexbackup.volume_header_info\n" && find . -xdev -depth ! -regex "news/articles" ! -regex ".*~"$ -print ) | afio -o -E /tmp/nocompress.981 -z -P gzip -Q -4 -Z -M 2m -T 3k -v -a -b 10k - | mbuffer -m 3M -s 10k | dd ibs=10k obs=10k conv=noerror,sync of="/dev/nst0"
[root@my-server root]$
Gah. Ok, so I guess I change it back to using 'dump' and see if it's still broken.
G
-
After changing back to using DUMP, and attempting a manual backup, I get :
[root@my-server root]$ /usr/bin/flexbackup -set all
flexbackup version 1.1.6 (http://flexbackup.sourceforge.net)
/etc/flexbackup.conf syntax OK
|------------------------------------------------------------
| Checking 'mbuffer' on this machine... Ok
| Checking /bin/sh on this machine... bash2
|------------------------------------------------------------
| Trying "mt defblksize" instead of "mt setblk"
| Doing level 0 backup of set all using dump
| All sets = full (1 tapes)
| Retensioning tape...
| Rewinding & erasing tape...
| Creating index key 200307231632
| Backup set full (/ /boot)
| Tape #0
|------------------------------------------------------------
At block 0.
|------------------------------------------------------------
| File number 1, tape index 200307231632
| Backup of: /
| NOTE: $exclude_expr is ignored for type=dump
| Date of this level 0 backup: Wed Jul 23 16:32:48 2003
| Date of last level 0 backup: the epoch
|------------------------------------------------------------
| dump -0 -b 10 -a -f - / | gzip -4 | mbuffer -m 3M -s 10k | dd ibs=10k
| obs=10k conv=noerror,sync of="/dev/nst0"
|------------------------------------------------------------
dd: opening /dev/nst0': Read-only file system
0.0 kB/s in - 0.0 kB/s out - 0 kB total - buffer 0% full DUMP: Date of this level 0 dump: Wed Jul 23 16:32:48 2003
DUMP: Dumping /dev/hde3 (/) to standard output
0.0 kB/s in - 0.0 kB/s out - 0 kB total - buffer 0% full DUMP: Added inode 8 to exclude list (journal inode)
DUMP: Added inode 7 to exclude list (resize inode)
DUMP: Label: /
DUMP: mapping (Pass I) [regular files]
0.0 kB/s in - 0.0 kB/s out - 0 kB total - buffer 0% full DUMP: mapping (Pass II) [directories]
DUMP: estimated 9150928 tape blocks.
DUMP: Volume 1 started with block 1 at: Wed Jul 23 16:33:33 2003
DUMP: dumping (Pass III) [directories]
DUMP: Broken pipe
DUMP: The ENTIRE dump is aborted.
ERROR from backup, exiting
offending command(s):
dump -0 -b 10 -a -f - / | gzip -4 | mbuffer -m 3M -s 10k | dd ibs=10k obs=10k conv=noerror,sync of="/dev/nst0"
I'm at a loss to even pinpoint where the error is here... this "Broken pipe" report... that's as meaningful to me as "fatal exception error".
G
-
Hi GP,
>dd: opening /dev/nst0': Read-only file system
I think there's your problem.
What tape drive are you using and is the system recognising it as st0 and nst0 ?
Kelvin
-
Kelvin wrote:
> >dd: opening /dev/nst0': Read-only file system
>
> I think there's your problem.
>
> What tape drive are you using and is the system recognising
> it as st0 and nst0 ?
And is your tape set to write-protected?
Charlie
-
Charlie Brady wrote:
>
> Kelvin wrote:
>
> > >dd: opening /dev/nst0': Read-only file system
> >
> > I think there's your problem.
> >
> > What tape drive are you using and is the system recognising
> > it as st0 and nst0 ?
It's a seagate Travan TR-5 10/20GB device. From dmesg :
hdb: Seagate STT20000A, ATAPI TAPE drive
...
scsi0 : SCSI host adapter emulation for IDE ATAPI devices
Vendor: Seagate Model: STT20000A Rev: 8A51
Type: Sequential-Access ANSI SCSI revision: 02
st: Version 20020205, bufsize 32768, wrt 30720, max init. bufs 4, s/g segs 16
Attached scsi tape st0 at scsi0, channel 0, id 0, lun 0
hda: DMA disabled
hdb: DMA disabled
> And is your tape set to write-protected?
No, it's definitely writeable...
G
-
Hi GP,
>It's a seagate Travan TR-5 10/20GB device.
Now, how did I just know you were going to say Travan :).
I remember reading on these phorums some time back about a strange problem with some Travans which require you to
1. insert the tape
2. run the backup command (which will fail)
3. then run the backup command again immediately after the first one has completed and failed, and the backup now runs
This may or may not apply to you, obviously, but perhaps worth a try ?
Another thing to try :-
1. Insert a tape ready for backup
2. Reboot the server with the tape already inserted.
3. Login and initiate a backup and see if this works.
Yet another thing to try :-
1. insert a tape
2. run tar -cvf /dev/st0 /home
Does tar complete without errors ?
Kelvin
-
Oops .. sorry, did not read the Subject line properly ....
Travan clearly mentioned ...... need new glasses :(
Kelvin
-
Kelvin wrote:
>
> Hi GP,
>
> >It's a seagate Travan TR-5 10/20GB device.
>
> Now, how did I just know you were going to say Travan :).
>
> I remember reading on these phorums some time back about a
> strange problem with some Travans which require you to
>
> 1. insert the tape
> 2. run the backup command (which will fail)
> 3. then run the backup command again immediately after the
> first one has completed and failed, and the backup now runs...
The other thing you really ought to check is whether the belt on top of the drive has come off. They are buggers for doing this, and they really don't work too well without it. Getting impatient and pulling a tape out at the wrong momnet will do it every time.
The best advice, however, is to chuck it in the trash and get something reliable and less difficult to set up.
Ed Form
-
Kelvin, thanks for the tips. Ed, thankfully (touch wood) the drive is still 'driving' with the rubber band on it. I thought even floppies didn't use rubber bands anymore. It's disquieting in this digital world.
As it turns out, trying to run flexbackup from the console helped much to figure out what was up. When trying to use this latest flexbackup version, and also use afio instead of dump or tar (so that at least some of the backup should be accessible in the case of error) several changes are needed to the flexbackup.conf (and templates of course) as detailed earlier.
I just checked my latest log file - it backed up all my filesystems!
Next project is to re-install and engage the other HDD to attempt software raid on this thing, then do a full restore... see how that affects the system and all it's installed rpm's... is there a particular command to pass on to e-smith after such a restore? Like "signal event post-restore something something"?
G
-
Hi GP,
>Kelvin, thanks for the tips.
No Worries ! :)
>Next project is to re-install and engage the other HDD to attempt software raid
And after trying it you want something with more bells and whistles (literally ! :)), you could try something like the Accusys ACS 7500 :-
http://www.quasarsoft.com.au/hardware/index.php?topic=storage
Cheers !
Kelvin
-
Kelvin wrote:
> And after trying it you want something with more bells and
> whistles (literally ! :)), you could try something like the
> Accusys ACS 7500 :-
Looks nice - I think something would have to break before a case could be made for such kit... a few $100s is it?
G