What does your backup stack look like?

[Guusggg]

Hi everybody,

Lately I've been bothered by the fear of our company burning down and taking all the data we collected in InvenTree so far with it. This raised a question to me (as I'm basically a hobbyist, helping someone start up), what do you guys use for reliable backups of InvenTree?

Myself, I've been thinking about making cloud storage available to friends and family, and I would store the exports of InvenTree along with an SQL dump to be able to restore it. Do I need to think of more things? I think the general rule of data storage is; the more places, the better.

Also do you guys automate InvenTree backups?

I really like to hear how you guys manage this!

Kind regards,

[eeintech]

At my company we have multiple bare metal installation of InvenTree, they are all backed up more or less in the same way:

• PostgresSQL dump => single TGZ file
• JSON dump (redundant to DB dump) using invoke export-records => single JSON file
• Archive media folder (for all the images, labels, etc.) => single TGZ file

Those 3 files are then backed up daily on our Git server, which is a separate server from where all instances are running (yes more places the better).
The VM running on this server is also backed-up weekly on a different location, which gives another backup option in case our Git server and VM fail at the same time.

Side note, I have tried restoring from both PostgresSQL dump and JSON dump and got the same result so it's probably not necessary to do both.

[matmair]

@eeintech I did a few test recoveries, JSON or Postgres is fine, you lose some admin-logs and stuff with postgres but nothing that is needed for InvenTree itself. JSON is good for the backup software I use so I chose that. Both seem redundant but that is kind of the point 😂

[matmair]

I have a cron job running that dumps the data as json and pushes it into a backup endpoint for veeam, the media path is set directly to a versioned fileshare. Also, the whole VM is replicated multiple times a day. In most orgs I deploy to they have systems on other floors / other buildings and cities as backup targets. But that gets managed by their IT, my personal InvenTree is only replicated to one DR.

[wolffam]

What's the suggested way to take a backup from a bare metal installation and ingest that backup into a newly provisioned Docker deployment of Inventree?

[wolffam]

I think we got it figured out. We were on step 6 pretty much. We had to remap some of the defaults in the config.yaml so that it was correctly pointing at the database container and it seems to be configured correctly now.

Thanks for thinking through it!

[wolffam]

The one remaining item I can't seem to figure out is whether or not there is a way to transfer user accounts from one deployment to another. Does anyone know? It's not a big deal to remake them but I figure I would ask first

[SchrodingersGat]

The user account data is stored in the same database, so it will be transferred across in the same process.

[matmair]

@wolffam if the migration process keeps presenting problems and you have time I would appreciate a short issue in the docs repo with what info is missing. This could become a common problem / question and I think we should address this in the docs.

[wolffam]

I think most of my issues were due to a lack of understanding of the docker process/compose in general. Once I got a handle on that, the current docs were helpful enough to guide me the rest of the way