Netbox 3.5 - script storage across multiple containers

[FliesLikeABrick]

Hello,

We have multiple containers running on different hosts for HA to our netbox setup. We previously managed our custom script deployment by tracking those files in our internal fork of netbox-docker, and exposing them to the container via a volume mapping from the local host (that had our netbox-docker fork cloned locally).

With 3.5's change that requires scripts to be uploaded via the UI, it does not look like this strategy works? Is there a recommended approach to synchronizing these reasonably? I would like to avoid needing a shared nfs volume or something for the script storage that the UI manages, if possible - since we already have our scripts versioned in git. Is there a way to just deploy these either in the container build or in the filesystem of the physical host?

[zmaster7]

Check out my response in #12663, which would be along the same lines. You'd be able to tie directly into your git repo and sync changes, but as for now the "upload" process is manual, even when a data source is involved. However, the file changes will be synced automatically when checked.

[abhi1693]

Have you tried using a local data source? If you shipped everything within the image, you could simply use the local backend to fetch your reports and scripts. It's just an additional step but nothing is required to be changed in your deployment. This also doesn't create anything extra on the filesystem.

[rasanentimo]

Can that be used also scripts/reports which are originated from local/git data source or is this only for manually uploaded files?

Having s3fs mount for the scripts directory could also work.

[abhi1693]

The idea of the fix is to keep the reports and scripts on S3 (if defined) irrespective of how they were synced

[rasanentimo]

ok, that sounds good! However, I guess there would be still the manual step of adding the scripts from the data source?

[abhi1693]

You can automate that too using initializers during the bootup process of your pod

[rasanentimo]

Actually, I mean if I add a new script/file to the git repo, I guess there's still a manual step to add new script(s) from the data source on an already running pod?