Topic: Problem with FlexBackup

[timm]

I'm checking to see if there is a real hardware issue, but If not there might be an issue with flexbackup.

Here is what I have:
Athlon XP2200
1Gig Ram
160 EIDE HD
SIS 760
SME 7.8b
Adaptec AHAXXXX SCSI card
Sony SDT-9000 DSS3 tape backup

Did a standard backup to tape.  Completed with no errors.  Then commanded a restore from tape..ended up with a broken mess. The screen changed to "internal server error"

Forced the server to re-configure and got it going again.  So then I start to check the tape drive with a known good backup from another SCO Unix server.  CPIO lists all files just perfect (from the SCO server)..  I attempt to list files with CPIO from flexbackup and I get no listing and then it spits out the tape with "end of file" listing.  I check the log from flexbackup and it appears to leave 2 seperate files on the tape? 1 index and then another with the files in it? Here is the log:

|------------------------------------------------------------
| Doing level 0 backup of set all using tar
| All sets = full (1 tapes)
| Rewinding & erasing tape...
| Found index key 200512181625, next file is 2
| Deleting record for 200512181625 file 0
| Deleting record for 200512181625 file 1
|------------------------------------------------------------
| Creating index key 200512181843
| Backup set "full" (/)
| Tape #0
|------------------------------------------------------------
At block 1.
|------------------------------------------------------------
| File number 1, tape index 200512181843
| Backup of: /
| Date of this level 0 backup: Sun Dec 18 18:43:40 2005
| Date of last level 0 backup: the epoch
|------------------------------------------------------------
| cd "/" && /tmp/collectexit.3898.sh find . -depth -xdev ! -type s ! -regex \
|   "news/articles" ! -regex ".*~"$ -print0 | /tmp/collectexit.3898.sh tar \
|   --create --null --files-from=- --ignore-failed-read --same-permissions \
|   --no-recursion --totals --label "level 0 / Sun Dec 18 18:43:40 2005 \
|   tar+gzip from bserver" --sparse --atime-preserve -b 64 --file - | \
|   /tmp/collectexit.3898.sh gzip -4 | /tmp/collectexit.3898.sh buffer -m \
|   20m -p 75 -s 32k -t -u 100 -B -o "/dev/nst0"
| [ ! -e /tmp/exitstatus.3898 ]
|------------------------------------------------------------
Total bytes written: 848625664 (810MiB, 3.3MiB/s)
Kilobytes Out 298720
|------------------------------------------------------------
| Backup start: Sun Dec 18 18:43:40 2005
| Backup end:   Sun Dec 18 18:48:08 2005
|------------------------------------------------------------
At block 9336.
|------------------------------------------------------------
| Rewinding...
| Removing old level 0 log of all (dated 200512181625)
|------------------------------------------------------------

I can access the tape drive through /dev/st0 or dev/nst0 or /dev/tape  it looks like flexbackup used /dev/nst0..

Should I attempt an uncompressed tar backup from /home, /user, /var and see if CPIO can list and restore all?

Where can I adjust the template for flexbackup to turn off the compression and perfrom other adjustments?

Please help.

-Tim

[CharlieBrady]

Did a standard backup to tape.  Completed with no errors.  Then commanded a restore from tape..ended up with a broken mess. The screen changed to "internal server error"

More details about the failure of the panel will be found in /var/log/httpd/admin_error_log.

[CharlieBrady]

I attempt to list files with CPIO from flexbackup and I get no listing and then it spits out the tape with "end of file" listing.

flexbackup in SME server doesn't use cpio, it uses tar (or dump, depending on version).

I check the log from flexbackup and it appears to leave 2 seperate files on the tape? 1 index and then another with the files in it?

That depends again on version. It might be three files; index, and contents of /boot and /.

I can access the tape drive through /dev/st0 or dev/nst0 or /dev/tape  it looks like flexbackup used /dev/nst0..

You need to use /dev/nst0, otherwise you can only access the first file on the tape - the index.

Where can I adjust the template for flexbackup to turn off the compression and perfrom other adjustments?

You'd need to create a custom template fragment for /etc/flexbackup.conf/10Compression. But that would be premature until you diagnose what the actual problem is.

Forced the server to re-configure and got it going again.

Restore should only be done onto a pristine freshly installed system. Otherwise you can't know what you end up with.

[timm]

Charlie,

Thanks for the quick reply.  I'm looking into what you stated. I found out that the jumper setting on the back of tape drive has a wierd ability to help keep the hardware compression on or off, So I'm making it so it is software settable in the hope that it helps.  I'll keep you posted later today.

-Tim

[timm]

OK I checked all the settings and made sure that the hardware compression on the tape is disabled by changing the jumpers. Still get the same problem.  It seems to destroy the server-manager panel and it won't work after it is done.

Here are the logs you spoke about:

[Wed Dec 21 09:13:26 2005] [error] [client 127.0.0.1] Error occurred restoring files from tape., referer: https://192.168.0.169/server-manager//cgi-bin/backup
[Wed Dec 21 09:13:36 2005] [error] [client 127.0.0.1] CANNOT_OPENconfiguration db at /etc/e-smith/web/panels/manager/cgi-bin/backup line 45., referer: https://192.168.0.169/server-manager
[Wed Dec 21 09:13:37 2005] [error] [client 127.0.0.1] Premature end of script headers: backup, referer: https://192.168.0.169/server-manager
[Wed Dec 21 09:14:00 2005] [error] [client 127.0.0.1] Can't call method "get" on an undefined value at /etc/e-smith/web/panels/manager/cgi-bin/viewlogfiles line 40., referer: https://192.168.0.169/server-manager
[Wed Dec 21 09:14:00 2005] [error] [client 127.0.0.1] Premature end of script headers: viewlogfiles, referer: https://192.168.0.169/server-manager
[Wed Dec 21 09:15:26 2005] [notice] caught SIGTERM, shutting down

I did check the hardware and a standard tar backup and cpio restore works fine...
 

To get the machine working after it's failed restore, I have to login as root and:

/sbin/e-smith/activate_config
reboot
login as admin and put back all of the configuration settings

"help"

-Tim

[CharlieBrady]

Still get the same problem.

I missed this bit of your original report.

 SME 7.8b

7.0beta8 is a beta *test* version. It's only out there so that you can find problems, and report them via the bug tracker.

If you look in the bug tracker, however, you'll find that this problem has already been reported:

http://bugs.contribs.org/show_bug.cgi?id=229

"Unable to restore from tape"

There is not yet a fix or a workaround, but if you keep checking that URL you will eventually find one.

[timm]

Still get the same problem.

I missed this bit of your original report.

 SME 7.8b

7.0beta8 is a beta *test* version. It's only out there so that you can find problems, and report them via the bug tracker.

If you look in the bug tracker, however, you'll find that this problem has already been reported:

http://bugs.contribs.org/show_bug.cgi?id=229

"Unable to restore from tape"

There is not yet a fix or a workaround, but if you keep checking that URL you will eventually find one.

Charlie,

Thanks for your help.  I was cautious about calling it a bug, since I had a thought that I may have been a hardware issue.  Thank you for your help.  I guess I should just do a standard tar backup until I know better.  What directories would you reccomend?  Should I use a dump command instead?  At the moment, my tape will hold all of the server.

Could I instead just replace the 7.8b template with a template for tar and use it instead?  If so where would I change this?  

Thanks for your assistance.

-Tim

[timm]

I found some of an answer to my own question.  I found an article at:

http://www-128.ibm.com/developerworks/linux/library/l-roadmap8/

Hope this helps..

-Tim