Opened 7 years ago

Closed 7 years ago

Last modified 6 years ago

#615 closed defect (fixed)

Mondo failed "No space left on device"

Reported by: cateyed Owned by: bruno
Priority: normal Milestone: 3.0.2
Component: mondo Version: 3.0.1
Severity: normal Keywords:
Cc:

Description

Hi there,

I'm still trying to backup (image) my system with mondo.

My command is "sudo mondoarchive -O -i -N -g -d /mnt/jeff/aras_backup -T

/tmp/mondo -S /usr/mondo/scratch -E "/media" -s 4480m

It now fails with "rm: cannot remove `/tmp/mondo/mondo.tmp.ap4XtT/mnt1': Device or resource busy"

Could you please advise how to proceed?

Many thanks,

Meaghan

Attachments (1)

mondo.zip (34.3 KB) - added by cateyed 7 years ago.

Download all attachments as: .zip

Change History (14)

Changed 7 years ago by cateyed

comment:1 Changed 7 years ago by bruno

The error in your log is:

cp: writing `/tmp/mondo/mondo.tmp.ap4XtT/mnt1/tmp/mountlist.txt': No space left on device
FATAL ERROR. Cannot copy mountlist to ramdisk

I don't see the error msg you mention.

Try increasing EXTRA_SPACE as per #598.

comment:2 Changed 7 years ago by cateyed

Hi again,

Sadly, this hasn't worked.

I have edited mindi.conf to have the following lines:

EXTRA_SPACE=92160
BOOT_SIZE=40960

The output on the screen is:

arasv@aras-linux ~ $ sudo mondoarchive -O -i -N -g -d /mnt/jeff/aras_backup -T /tmp/mondo -S /usr/mondo/scratch -E "/media" -s 4480m 
Initializing...                                                                                                                      
df: `/tmp/mondo/mondo.tmp.zZ8jeb/mnt1': No such file or directory                                                                    
df: `/tmp/mondo/mondo.tmp.XSTkCE/mnt1': No such file or directory                                                                    
See /var/log/mondoarchive.log for details of backup run.                                                                             
Checking sanity of your Linux distribution                                                                                           
Done.                                                                                                                                
---FATALERROR--- Failed to generate boot+data disks                                                                                  
Execution run ended; result=254t, please contact the mailing list.                                                                   
Type 'less /var/log/mondoarchive.log' to see the output log                                                                          
rm: cannot remove `/tmp/mondo/mondo.tmp.jE3KMR/mnt1': Device or resource busy                   


mindi.log has the same error as before:

cp: writing `/tmp/mondo/mondo.tmp.jE3KMR/mnt1/tmp/mountlist.txt': No space left on device
FATAL ERROR. Cannot copy mountlist to ramdisk

comment:3 Changed 7 years ago by cateyed

  • Summary changed from Mondo failed "Device or resource busy" to Mondo failed "No space left on device"

comment:4 follow-up: Changed 7 years ago by cateyed

I just did a df -h whilst mondo was running, at the point that it stops and I get this result:

No space left on device

root@aras-linux scratch # df -h
Filesystem            Size  Used Avail Use% Mounted on
/dev/sda1             547G   95G  424G  19% /
none                  2.0G  716K  2.0G   1% /dev
none                  2.0G  1.6M  2.0G   1% /dev/shm
none                  2.0G  128K  2.0G   1% /var/run
none                  2.0G     0  2.0G   0% /var/lock
ocean:/var/lib/samba/share/development/40_Magellan_Standard_Software/MagTiny/v3
                      188G  144G   34G  82% /scratchbox/users/arasv/home/arasv/work
ocean:/var/lib/samba/share
                      188G  144G   34G  82% /samba
ocean:/home           192G  161G   32G  84% /home_ocean
192.168.65.205:/scratchbox
                       96G   12G   79G  14% /mnt/jeff
df: `/tmp/mondo/mondo.tmp.zZ8jeb/mnt1': No such file or directory
df: `/tmp/mondo/mondo.tmp.XSTkCE/mnt1': No such file or directory
/dev/loop2            128M  128M     0 100% /tmp/mondo/mondo.tmp.srdhxb/mnt1

Note that the mount point /dev/loop2, which is the mondo tmp directory is showing as 100% full. Could this be the issue?

Once mondo fails, the mount point fails to be umounted, here it is now:

/dev/loop2            128M  1.6M  125M   2% /tmp/mondo/mondo.tmp.srdhxb/mnt1

Sorry if I'm off on a tangent.... Just trying to work this out myself! :-)

comment:5 in reply to: ↑ 4 Changed 7 years ago by bruno

  • Status changed from new to assigned

Replying to cateyed:

Note that the mount point /dev/loop2, which is the mondo tmp directory is showing as 100% full. Could this be the issue?

yes this is !

Sorry if I'm off on a tangent.... Just trying to work this out myself! :-)

Which is great !! So I think your issue is still related to a size problem. Can you change the 92160 into 256000 just to check ?

What is your distribution ?

Also be sure to unmount remaining loop FS before relaunching.

comment:6 Changed 7 years ago by cateyed

Hi again,

Sadly, same result.

I modified /etc/mindi/mindi.conf and changed as suggested above.

I modified /etc/mtab and hashed out the old mount points that couldn't be umounted.

Re-ran the command and df still shows:

/dev/loop3            278M  278M     0 100% /tmp/mondo/mondo.tmp.6DG59R/mnt1

We are running Ubuntu 11.04 64 bit.

Many thanks for all your help so far!

comment:7 Changed 7 years ago by cateyed

Hi there,

Is there any further help you can provide on this issue? Or will I have to give up?

Thanks,

Meaghan

comment:8 Changed 7 years ago by bruno

Will you be able to test the new beta of 3.0.2 I've launched today ? Will be shortly available under ftp://ftp.mondorescue.org/test/ubuntu/11.04/

This is reverting a patch which seemed to cause the larger use of the temp mounted dir. So hopefully you won't get the same issue as before.

comment:9 Changed 7 years ago by cateyed

Hi Bruno,

Thanks for the new release. Sadly it didn't seem to make any difference.

Meaghan

comment:10 Changed 7 years ago by bruno

I'm not really seeing any good reason for your error as of now. Would you mind trying to revert to an earlier version ? (2.2.9.7 would be probably a good previous one - with mindi 2.0.7.8) If that one works, I could look at the diff to see what could lead to your issue.

comment:11 Changed 7 years ago by bruno

Probably linked to #617 as well.

comment:12 Changed 7 years ago by bruno

  • Resolution set to fixed
  • Status changed from assigned to closed

Fixed in [3014].

Analysis of deps returns /lib64 on Ubuntu and Debian as it's a link, so needed in the rootfs, and thus ReadAllLink? expands it to /lib. But including a full /lib makes stuf behaving as reported here and in #617.

So now real directories are excluded from the tar transfer to the rootfs in order to avoid that.

Mindi now works on Ubuntu 11.04 x86_64

Last edited 7 years ago by bruno (previous) (diff)

comment:13 Changed 6 years ago by bruno

Also look at #689 for another related issue fixed. Planned to be part of mindi 2.1.6

Note: See TracTickets for help on using tickets.