[Solved] Do first image backup on local server then move image to internet server for subsequent incremental image backups

Hi

I’m in love with urbackup. I’m using it for LAN backups with very good results.

I wonder if this idea can be accomplished.

I have several offices that I want to backup on an internet urbackup server. However the process of uploading the first full image of each server is too tedious and slow (300GB can be more than 18 days with our regular spanish dsl bandwidth). Since we have physical access to our internet server I wonder if I can use on each office a local urbackup server to do the first full image and then move that image and its configuration to the main internet server by copying the files. Then do the subsequent incremental backups through the WAN.

I’ve already seen in other threads similar questions but unanswered:

If this can be done it would be of great help to know which files aside of the vhdz image should I move between servers. Or maybe which table records should I export from local urbackup server to internet urbackup server

Thank you in advance

1 Like

Seed the backup, this wont work with vhd :

You already run a local backup server on each office
Take an external hdd, copy all the backuped files in bulk to it ( wont really matter all you need is the file and hash)…
Look at the forums/faq for the key files to copy and where (4-6 of them).
Easiest is to use the same keys on all server, other option is to have different server keys and on the client put the keys of all servers.
Maybe check that locally your clients still work with this set of keys
Make a dumb client on the WAN , make one backup using the HDD attached to it.
Make one backup of all the real client (from now it will transfert only hashes)
Then you can delete the dumb client

Theses instructions were already given in a way or another in the forums.

Thank you orogor for your quick answer.

I understand that you are talking about file backup. I already know how to do this with urbackup. My question is about image backups.

I wonder if I push a local urbackup vhdz generated image to the internet server, the internet server would be able to do the subsequent incremental image backups.

I guess I’d need to upload also the database entries where the image hashes are stored. So that was my question. Can it be done? If so what database files should I look at?

Thank you in advance.

Can you confirm that there is no way to populate a internet server with a VHD image obtained in a localserver (so the internet server uses it for incremental images)?

Thanks.

Hi,

I have found your post and I have the following idea.
You could set up the backup server in a virtual machine (running locally in your office), do the full image backup and then copy the whole virtual machine on your internet server and start it from there.

This is just an idea and I have not tested it before - but I am looking for an identical solution right now.

Kind regards,
Wulffi

Hi Wulffi007,

that’s a cool idea, however I have several offices and one internet backup server, so in my case I cannot have one VM for each office.

I’m doing some work on this stuff, when I get some positive results I’ll post them here.

One more advanced option would be to manually fiddle with the backup_server.db database (e.g. with Database Browser for SQLite).

Add the client. Note the id in the clients table. Then copy the rows in backup_images and assoc_images adjusting the ids.

Thanks uroni, I’ll try your hint.

I’ll schedule a test at the end of this week or at the beginning of next week. I’ll come back with feedback.

Hi uroni

I’m urhmontoliu’s colleague in this issue. I prepared sql scripts for feed tables backup_images and assoc_images. We have a doubt: when a client has more than 2 images (i.e. drives C: and D:), last image is not registered in assoc_images table.

backup_images:

"54","14","2017-02-17 10:44:19","0","0","/var/local/backup/urbackup/ClientName/170217-1144_Image_SYSVOL/Image_SYSVOL_170217-1144.vhdz","1","2017-02-17 10:44:53","336437248","1","SYSVOL","1487328293","0",,,
"56","14","2017-02-17 10:44:54","0","0","/var/local/backup/urbackup/ClientName/170217-1144_Image_C/Image_C_170217-1144.vhdz","1","2017-02-17 14:39:19","86104014848","1","C:","1487342359","0",,,
"61","14","2017-02-20 12:14:19","0","0","/var/local/backup/urbackup/ClientName/170220-1314_Image_D/Image_D_170220-1314.vhdz","1","2017-02-20 12:25:36","5140750336","1","D:","1487593536","0",,,

assoc_images:

"56","54"

Can you explain this for us?

Thanks in advance

We have a working solution for migrating IMAGE backups from “temporary” local urbackup server to feed an internet server.

It also works for restoring images in low bandwidth environments.

You’ll need:

  • An USB disk
  • A Car
  • A temporary urbackup server

To feed the internet server, the process is:

  1. create a client for the internet server but avoid doing any backup

  2. use a local (temporal) urbackup server to create the first image (with 1 and 2 done you are sure that both local and internet servers have the same client name)

  3. copy the client image files tree and a copy of server_backup.db* data base files to an USB disk

  4. travel to the ISP where lives the internet server

  5. Mount the usb disk to the internet urbackup server and run the script to import the backup images (link below). The script will do

    • stop urbackup server for security
    • backup the production urbackup db for security
    • move the image files to its final location
    • feed the internet urbackup server data base with the right ids (changing them as needed)
    • start the urbackup server after all the operation is done

The same script also works for restoring, the process is:

  1. Travel to the ISP where the internet server lives,
  2. copy both the full and incremental images of the client to be restored and the server_backup.db* files to an USB disk
  3. travel to the location where the computer to be fully restored is. Take with you a temporary urbackup server.
  4. feed the temporary urbackup server database and backup image folders using the script (link below).
  5. restore the computer with the live CD using as restore server the temporary one.

We have tested the procedure in both ways.

Use at your own risk. Code can be found at

HTH

1 Like

Just a question for uroni

we are considering that the backup_images table has it’s own independent primary key IDs, however we have some doubts. In some of our urbackup servers seems that both backup_images and backups tables increment their primary key IDs in the same way. Are the backup_images’s ids completely independent?

Thank you

Yes, the primary keys are independent. One can also insert without specifying the primary key and SQLite will create one automatically.

Thank you uroni