Restoring project after deletion

MRvLuijpen
Restoring project after deletion

Hello Community.

I have a question, one of the users deleted a project (without creating a backup) and I am now curious if this project can be restored. We do have a backup available, but of course would only like to restore this one project (and not damaging any of the other active & running projects).

Which folders should we restore and is restoring those folders enough for the project to be restored.

probably part of the folders:

  • dss_data/timelines
  • dss_data/scenarios
  • dss_data/config/projects
  • dss_data/managed_folders
  • dss_data/managed_datasets

Thank you in advance,

Marc Robert

0 Kudos
5 Replies
ATsao
Dataiker

Hi Marc, 

The cleanest way would by far be to restore the backup on a separate instance or location, export the target project, and then import said project into your original instance. Then, you could delete the backup instance if it's no longer needed. 

Best,

Andrew

0 Kudos
MRvLuijpen
Author
Thank you Andrew @ATsao
0 Kudos
MRvLuijpen
Author

Hello Andrew ( @ATsao)

It seems that restoring the previous version is not easy to do, since it is the production machine, and restoring the older version would have major impact on all other users.

Is there a way to retrieve it through the internal GIT of dataiku?

Thanks again.

Marc Robert

0 Kudos
ATsao
Dataiker

Hi Marc, 

If you have a backup of the data directory, you can simply restore it elsewhere (such as a different machine), export the project, and then reimport back into your production instance. You'll want to follow the steps that have been outlined here: 

https://doc.dataiku.com/dss/latest/operations/backups.html#restore-on-another-machine-another-locati...

You won't be able to restore the project via Git as Git only stores the history related to the DSS configuration. The alternative option is to simply copy/paste the deleted project from the backup location into your <DATA_DIR>/config/projects folder while DSS is stopped. Please note that this would only simply restore the Flow and recipes itself, not a full restore of the project, as there will be a lot of information that is stored in other areas of the data directory (for example, uploaded data, contents of your managed folder, model data, notebooks, etc). This is why we strongly recommend restoring your data directory separately and doing an export/import of the project, which will preserve these additional references when the project is imported back into DSS. 

Best,

Andrew

0 Kudos
pvannies

Dear Community,

This is for other community users ending up here.
If you have an up-to-date remote git coupled to your Dataiku project, and you accidentally deleted your project and it is not possible anymore to get a backup version as described above, please not that you can restore your project configurations using the remote git. As @ATsao mentioned, it will not restore uploaded data, etc. But it will restore your flow, which is already quite helpful. This is how to do it:

  • create a new project inside Dataiku.
  • link it to an empty remote git repository and push once (necessary to avoid errors upon first pull).
  • now edit the remote to the repository of the project you want to restore.
  • perform a pull operation to restore your old project configurations.

Hope this helps someone!
Cheers Pauline