Created attachment 717722 [details]
File: backtrace

Created attachment 717723 [details]
File: core_backtrace

Created attachment 717724 [details]
File: environ

Created attachment 717725 [details]
File: smolt_data

So, this happens when your backup dir is messed up. ;) 

Can you attach the output of: 

ls -la /mnt/backup/boot/rdiff-backup-data/current_mirror*

You should normally have one of those showing the current backup. 
If you have 2 of them it means a backup failed and needs to regress. 
If you have more than 2, you get this error and something tampered with your backup dir somehow.

ls -la /mnt/backup/boot/rdiff-backup-data/current_mirror*

-rw-------. 1 root root 10 Mar 15 01:06 /mnt/backup/boot/rdiff-backup-data/current_mirror.2013-03-15T01:06:02-05:00.data
-rw-------. 1 root root 10 Mar 15 01:08 /mnt/backup/boot/rdiff-backup-data/current_mirror.2013-03-15T01:07:02-05:00.data
-rw-------. 1 root root 10 Mar 15 01:08 /mnt/backup/boot/rdiff-backup-data/current_mirror.2013-03-15T01:08:02-05:00.data

I've deleted items like this before, but it keeps occurring so I thought to submit this. I'm tempted just to format the external disk and move on, but have seen enough google hits to thing something else is going on.

Thanks for the quick response. I'm not sure where to start debugging this.

Here's my bash script run at 1 AM.

#!/bin/bash
mount | grep -q /mnt/backup
if [ $? -eq 0 ]
then
   for i in /boot /etc /var/log /var/account /home /root /opt
   do
      echo /usr/bin/rdiff-backup --print-statistics ${i} /mnt/backup${i}
      /usr/bin/rdiff-backup --print-statistics ${i} /mnt/backup${i}
   done
else
   echo "External disk NOT mounted"
fi | tee /root/backup.log

It seems like it's running 3 times? once every minute?

Check your cron log?

There's a huge amount of runs happening. I'm not sure why. 

I've wrapped the script with some PID file magic to try and prevent multiple instances from running.

# grep rdiff-backup /var/log/cron
Apr  1 01:00:01 CROND[6226]: (root) CMD (/root/bin/rdiff-backup)
...
Apr  1 01:59:01 CROND[6226]: (root) CMD (/root/bin/rdiff-backup)

# grep rdiff-backup /var/log/cron | wc -l
60

It's running my script every minute for 1 hour.

I believe the answer is rdiff-backup may be running from the previous day if there is enough data churn. Meaning, the current day's entry will start to run creating an overlapping situation since the previous day's run is still happening. The current day's instance creates the multiple current_mirror.* file and once that happens the script will always die horribly. cron dutifully detects the failure and retries every minute until the next hour.

Does this make sense?

Maybe, rdiff-backup shouldn't die so horribly when it detects too many current_mirror.* files and provide an easier to decipher error message? A nicety, not a necessity...

If my diagnoses makes sense then excellent. If you think it's something else then let me know what you need.

"I've wrapped the script with some PID file magic to try and prevent multiple instances from running."

Sorry, meaning that I just implemented that right now, today.

Sure, that sounds plausable. 

I guess lets see if your fix works tonight?

Please close this issue.

PEBKAC... :(

Sure. It happens to us all... no problem.

See also: bug 1109563