Koozali.org: home of the SME Server

Consistent flexbackup failures

Des Dougan

Consistent flexbackup failures
« on: March 25, 2002, 07:32:25 AM »
I'm having a consistent, reproducible failure with flexbackup. Before I report it to the bugs address, I'd like some feedback on whether others have seen this issue with 5.1.2.

I have a new server, with a SCSI Seagate Travan 5 drive. Each time backup runs _from cron_ it fails:


|------------------------------------------------
| Doing level 0 backup of all using dump
| Retensioning tape...
| Rewinding & erasing tape...
| Creating index key 200203241913.01
| Tape #0
| Filesystems = / /boot
|------------------------------------------------
At block 1.
|------------------------------------------------
| File number 1, index key 200203241913.01
| Backup of: /
| Date of this level 0 backup: Sun Mar 24 19:13:04 2002
| Date of last level 0 backup: the epoch
|------------------------------------------------
| (dump -0 -b 10 -a -f - / | gzip -4) | buffer -m 3m -s 10k -u 100 -t -p 75 -B \
|  -o /dev/nst0
|------------------------------------------------
  DUMP: Date of this level 0 dump: Sun Mar 24 19:13:04 2002
  DUMP: Date of last level 0 dump: the epoch
  DUMP: Dumping /dev/md1 (/) to standard output
  DUMP: Label: none
  DUMP: mapping (Pass I) [regular files]
  DUMP: mapping (Pass II) [directories]
  DUMP: estimated 3019549 tape blocks.
  DUMP: Volume 1 started at: Sun Mar 24 19:15:33 2002
  DUMP: dumping (Pass III) [directories]
  DUMP: dumping (Pass IV) [regular files]
buffer (writer): write of data failed: Input/output error
bytes to write=10240, bytes written=-1, total written         10K
  DUMP: Broken pipe
  DUMP: The ENTIRE dump is aborted.

If I run the backup command from the command line without removing the tape, the backup works. I have proven to my satisfaction that the problem is related to the tape being removed from the drive. If the tape is removed and replaced before running from the command line, it will fail again.

It would appear that the problem is something to do with the tape header, and, if I recall correctly, there was a problem of this nature with version 4.1.

I would appreciate any feedback on this.

Thanks,

Des Dougan

Rob

Re: Consistent flexbackup failures
« Reply #1 on: March 25, 2002, 08:50:36 AM »
A question

If the tape stays in the machine you can backup from command line. But if you take last nights tape out and put in todays it won't run?  Is that right?

Rob

Des Dougan

Re: Consistent flexbackup failures
« Reply #2 on: March 25, 2002, 08:56:10 AM »
Rob,

Yes, that's correct. Today's won't run the first time, but will if I run the command a second time without taking the tape out.

Des

Rob

Re: Consistent flexbackup failures
« Reply #3 on: March 25, 2002, 09:29:43 AM »
Hmm..

Have you tried doing a test on version 5.0?

Darrell May

Re: Consistent flexbackup failures
« Reply #4 on: March 25, 2002, 11:14:38 AM »
Des Dougan wrote:
> |------------------------------------------------
> | Doing level 0 backup of all using dump
> | Retensioning tape...
> | Rewinding & erasing tape...
> | Creating index key 200203241913.01
> | Tape #0
> | Filesystems = / /boot
> |------------------------------------------------
> At block 1.

Your error is right here.  At block 1 means you are failing to write the initial 32k block at the beginning of your backup.  In /sbin/e-smith/backup you will find this:

/bin/dd if=/dev/zero of="/dev/nst0" bs=32k count=1 > /dev/null 2>&1

This should position the backup to start at a block far after 1.  On my usage I start at block 85.

So I believe what you are finding is a tape that is new or has been erased is failing but a tape that has been written to and not erased works.

Anyhow this should give you an idea to troubleshoot further.

Regards,

Des Dougan

Re: Consistent flexbackup failures
« Reply #5 on: March 30, 2002, 09:04:52 AM »
Darrell,

Thanks for your suggestion. I created a fragment and changed the block count as you'd suggested:

    /bin/dd if=/dev/zero of="/dev/nst0" bs=32k count=85 > /dev/null 2>&1

However, I'm still getting the same error, and it still indicates block 1:

[root@jeeves backup]# /sbin/e-smith/backup

flexbackup version 0.9.8
/etc/flexbackup.conf syntax OK

|------------------------------------------------
| Doing level 0 backup of all using dump
| Retensioning tape...
| Rewinding & erasing tape...
| Creating index key 200203292154.42
| Tape #0
| Filesystems = / /boot
|------------------------------------------------
At block 1.
|------------------------------------------------
| File number 1, index key 200203292154.42
| Backup of: /
| Date of this level 0 backup: Fri Mar 29 21:54:44 2002
| Date of last level 0 backup: the epoch
|------------------------------------------------
| (dump -0 -b 10 -a -f - / | gzip -4) | buffer -m 3m -s 10k -u 100 -t -p 75 -B \
|  -o /dev/nst0
|------------------------------------------------
  DUMP: Date of this level 0 dump: Fri Mar 29 21:54:44 2002
  DUMP: Date of last level 0 dump: the epoch
  DUMP: Dumping /dev/md1 (/) to standard output
  DUMP: Label: none
  DUMP: mapping (Pass I) [regular files]
  DUMP: mapping (Pass II) [directories]
  DUMP: estimated 3168389 tape blocks.
  DUMP: Volume 1 started at: Fri Mar 29 21:57:13 2002
  DUMP: dumping (Pass III) [directories]
  DUMP: dumping (Pass IV) [regular files]
buffer (writer): write of data failed: Input/output error
bytes to write=10240, bytes written=-1, total written         10K
  DUMP: Broken pipe
  DUMP: The ENTIRE dump is aborted.
|------------------------------------------------
| Backup start: Fri Mar 29 21:54:44 2002
| Backup end:   Fri Mar 29 21:57:51 2002
|------------------------------------------------
At block 1.
|------------------------------------------------
| File number 2, index key 200203292154.42
| Backup of: /boot
| Date of this level 0 backup: Fri Mar 29 21:58:01 2002
| Date of last level 0 backup: the epoch
|------------------------------------------------
| (dump -0 -b 10 -a -f - /boot | gzip -4) | buffer -m 3m -s 10k -u 100 -t -p \
|  75 -B -o /dev/nst0
|------------------------------------------------
  DUMP: Date of this level 0 dump: Fri Mar 29 21:58:01 2002
  DUMP: Date of last level 0 dump: the epoch
  DUMP: Dumping /dev/md0 (/boot) to standard output
  DUMP: Label: none
  DUMP: mapping (Pass I) [regular files]
  DUMP: mapping (Pass II) [directories]
  DUMP: estimated 5979 tape blocks.
  DUMP: Volume 1 started at: Fri Mar 29 21:58:02 2002
  DUMP: dumping (Pass III) [directories]
  DUMP: dumping (Pass IV) [regular files]
buffer (writer): write of data failed: Input/output error
bytes to write=10240, bytes written=-1, total written          0K
  DUMP: Broken pipe
  DUMP: The ENTIRE dump is aborted.
|------------------------------------------------
| Backup start: Fri Mar 29 21:58:01 2002
| Backup end:   Fri Mar 29 21:58:04 2002
|------------------------------------------------
At block 1.
|------------------------------------------------
| Rewinding...
| Compressing log (all.0.20020329.gz)
| Linking all.latest.gz -> all.0.20020329.gz
|------------------------------------------------

File  Contents    (tape index 200203292154.42)
-----------------------------------------------
0  
1   level 0 / Fri Mar 29 21:54:44 2002 dump+gzip from jeeves
2   level 0 /boot Fri Mar 29 21:58:01 2002 dump+gzip from jeeves

Given that it's been working successfully if the command is run again without removing the tape, could it be related to the tape heads positioning the start of tape marker?


Des Dougan

Jason

Re: Consistent flexbackup failures
« Reply #6 on: May 07, 2002, 11:24:59 PM »
This is a software bug in SME.  I have the exact same problem.  It is NOT a hardware issue.