supabase db reset --linked potentially create orphaned storage objects

[IPJT]

Hi there,

Previously (not sure what version) when running supabase db reset --linked it failed if there were storage objects and I had to manually remove the storage objects before successfully resetting the remote db.

On v2.15.8 (and most likely versions before that) it seems like supabase db reset --linked will wipe all the metadata but not the storage objects themselves and thus creating orphaned storage objects. Doing exactly what the documentation is warning about:

Deleting objects should always be done via the Storage API and NOT via a SQL query. Deleting objects via a SQL query will not remove the object from the bucket and will result in the object being orphaned.

[doried-a-a]

Similar issue here on v2.12.1, doing supabase db reset --linked is erasing the whole storage metadata.
I was not expecting performing this command to affect the storage schema at all, is it a bug?

[sweatybridge]

Resetting the storage schema is indeed a new behaviour we've shipped. The main use case is to more easily reset the project for seeding.

We are also trying out a new clean up job to remove orphaned objects so that wouldn't be an issue in the long term.

Could you elaborate why this behaviour is problematic for your use case? We can look into reverting this change and add a new flag instead for those who want it.

[IPJT]

@sweatybridge -
Happy to hear that you're working on a long term solution.
Not problematic for me as I was aware of this and deleted all the storage objects before resetting the remote DB. However, would have been very frustrating doing it the other way around with no possibility to delete the orphaned storage objects which would still account to my Supabase bill.

The docs clearly state the danger of trying to delete storage objects by tampering with the storage schema. As a user I'd taker it for granted that the officially supported CLI does not fall into that trap (which it kind of does right now).

Maybe a warning or something to mitigate this?

Cheers!

[arunegiwebmark]

don't know whether this issue has been fixed or not, i really wanted the storage to be clear on db reset, because in case of seeding it always errors.

[sweatybridge]

it's usually more flexible to use supabase seed buckets command with buckets declared in config.toml file. https://github.com/orgs/supabase/discussions/18937#discussioncomment-13984081

[Lucas-Irvine]

@sweatybridge Is there a way to full reset the storage buckets though? Using supabase seed buckets --linked adds data to my buckets, but it doesn't actually delete the files that are already in the prod buckets.