Topic: Restoring a dar backup on a clean installation

[Mattj]

Hi all,

Im trying to restore a backup from our main sme server on a clean install,

i install sme 8, run all updates, mount the removable disk so it is visible and i can view set1 set2 and set3.

however, when in the console in the backup and restore option, it states it cannot find the removable disk, /media/nolabel is not available, i have set the mount to be the same name as this and still it will not find my backups.

to get round this, i restored the /home/e-smith/db from one of the sets of my backup. using dar -x /media/backupdisk/"severname"/set1/fullxxxxxxxx -N -R / -w -g home/e-smith/db,

this allows me to view my backups in the console and select one of the sets, however when i try and restore, it process through then spits out the error, journal commit I/O error.

I understand that the way to get the console to see my backup is not the most ideal way, and very likely to be the wrong way.

any help or advice on the error i am receiving would be a huge help.

Kind Regards,

[Stefano]

journal commit I/O error.

sounds like a defective hd..
can you post the whole error message?

TIA

[janet]

Mattj

How was the backup originally performed ?
ie in server manager backup & restore panel ? selecting backup to desktop, or backup to workstation, to local USB drive or what ?
Did you perform the backup in the admin console (not server manager) ie login as admin & select backup now from the text style menu to a locally connected USB.
Did you use a backup contrib (eg DAR2) or the default backup & restore panel in server manager (ie backup to workstation) ?

There are different types of backups that can only be restored the appropriate way ie restore from server manager or restore at first boot after OS install.

What version of sme server did you perform the backup on (ie your main server) ?

Is the backup drive a SATA drive or a USB drive.

Typically the backup drive is not mounted, & the restore process will mount the drive & then do the restore.
It may give an error if already mounted.

Having to use non standard procedures is not good & usually indicates you are doing something wrong.

Also take a look at the Backup server config Howto for useful additional information & explanations, it may help you.
You should not need to use non standard approaches, so I recommend you answer my questions above & see if we can resolve your problem rather than finding workarounds.
Using workarounds will most probably result in a incorrectly restored server.

Remember if you are repeating the restore, then it should always & ONLY be to a freshly installed (clean) operating system, which has NOT had previous restores performed on it, you will end up with lots of inconsistencies in your data & configuration if you do repeated full restores on the same already used or "restored to" OS.
So reinstall SME OS before each new restore attempt.

[Mattj]

How was the backup originally performed ?

The backup is performed from the server-manager on a scheduled basis to an external sata HDD, via a usb caddy.

Here is the details on the backup configuration on the main server,

Backup is Enabled.
Backup is made on local removable disk
Destination backup share folder is media/backupdisk
Number of rotating backup sets is 3
Number of daily backups contained in each set is 1
Compression level (0-9) of backup is 6
Daily backup occurs at 17:30
Each daily backup session is cleanly timed out after 24 hours.except full backups which are cleanly timed out after 24 hours
Full backup sessions (new backup set) are allowed everyday

the version of SME that the backup is performed on is 8.1,
I am using an SME Server 8 disc, and running updates to be 8.1.

I am currently re-installing SME, and will attempt to use the console to restore again. i will post on what errors/issues i face.

thanks for the feedback

[Mattj]

sounds like a defective hd..
can you post the whole error message?

TIA

i only saw that error pop up, if it occurs again (i have currently re-installed the server), i will check the log for a more detailed message, i suspected a defect HDD too, but i ran some checks using chkdisk and some utilities on Hirens and it came back ok.

[janet]

Mattj

The backup is performed from the server-manager on a scheduled basis to an external sata HDD, via a usb caddy.
.....Destination backup share folder is media/backupdisk

.....will attempt to use the console to restore again.

So you made the backup using server manager, which creates a *.dar file or a series of *.dar files

In sme server parlance, the console is the admin console that you access when you login at the command prompt as admin.
Backups can be created from the text style graphical menu, but these create a smeserver.tgz file.

When sme server boots the first time after installing a new OS, you get asked if you want to restore from USB, & then get asked to connect the USB drive to a USB port.
This console restore process expects to see a backup file that was created using the console menu, which creates a smeserver.tgz file.

You cannot use a backup with *.dar files on it to restore from on first boot.
You have to restore *.dar files from the server manager backup & restore panel.

Does that possibly answer what your problem is ie following an incorrect procedure.

Edit
One other comment, did you Verify the backup after it was completed, if the backup is incomplete or faulty then the Verify option should show this.
Wait till it finishes the Verification, which can take a while.

[Mattj]

Unfortunately no,

I am aware i have to use the console to restore the .dar backups, my issue mainly is getting the backup to be visible while in the backup and restore option on the clean installation.

i have done the following so far;

installed a clean version of sme,

ran yum update, performed the signal-event ; reboot option it provides (not the exact syntax of course..),

logged in as root, ran fdisk -l to find my mount point for the local removable disk /dev/sdg1,

i added this into etc/fstab with the mount option of /media/backupdisk.

i then ran mkdir /media/backupdisk

so that i can view the files on the local removable disk.

i am now going into the console to see if i can restore, this is where i usually get the error in regards to /media/nolabel is not a mount point.

regards,

[Mattj]

Ok,

so i configured the workstation backup options and it shows the correct directory of /media/backupdisk, as opposed to the /media/nolabel i was experiencing in previous attempts,

however when i try and restore i am getting this message,

No directory for your host in shared folder. Maybe your host name is different from backup ones.

any ideas?

[janet]

Mattj

The backup drive should NOT be mounted when you start the restore process.
The drive will be mounted by the restore procedure.

So unmount the drive before you try the restore.

You should not need to put anything in fstab, sme should automatically find & mount the drive in /media/...labelname...
 
Can you try a standard USB drive that connects directly to a USB port, not a SATA in a USB caddie, maybe that combination has some issues being auto mounted

[Mattj]

i removed the entry in the fstab, rebooted and it auto mounts using the usb and hdd caddy, however i still receive the same message when trying to do a restore,

No directory for you host in shared folder. Maybe your host name is different from backup ones,

The removable disk is always unmounted, i'm aware restores mount the drives to access it, and if it is previously mounted it does not work.

Do i have to set the details/configuration of this server to be identical to the backup server?

i.e. same IP, server name etc..

It slipped my mind to mention, that i am restoring onto an offline machine, i'm doing this simply for backup verification to ensure that our backups are doing what they should be, evidently it is not as straight forward as i'd hoped.

[brianr]

No directory for your host in shared folder. Maybe your host name is different from backup ones.

Are you sure that the newly installed system has exactly the same primary domain name and hostname as the backup up system?  The directory on the external drive that DAR creates uses this full domain name and hostname of the server (e.g. myhost.mydomain.com). I have got around this in the past by renaming the directory to conform to the newly installed system.

[Mattj]

wow, i had missed a "." out of the host name..... *facepalm*,

it now sees the drive and attempting the restore.

is there anything other than HD failure that can cause the Journal Commit I/O Error?

Thanks everyone for the advice and help

[Stefano]

search google with "Journal Commit I/O Error" keys

HTH

[Mattj]

yeah that helps thanks, i think one of our main server HDD's may be going faulty, or our raid controller is on its way out.

i have solved my original problem as i can see the drive i want to restore, its just the physical restore doesn't work due to i/o errors which will have to be troubleshooted further.

thanks everyone