el-cacahueto

I have a solution for this related to supabase, your dms are closed so dm me if interested

You could use RLS to restrict crud operations in the db though you could also restrict it in your code.

Could you dm me too pls ?

No it's better for having a single project hosted on several servers

What about folders ?

That would be impossible as the services needs to be restarted for the changes to take effect.

Why not an automatic rag pipeline that would make it easy to add or query vectors from text inputs

You get the shirt code when results are disclosed, It's written at the end of the survey.... (only if you gave out you email btw)

It's a bit tricky but double/tripple clicking on the left side of the text or a bit on the right of the last line select the text most of the time. You can then copy it as long as you keep your mouse on it to prevent it from disappearing.

Not yet, i'm building a supabase drop in replacement alternative and am looking to add more providers out of the box like QQ so i'll keep you updated on my experience with it

Use the custom providers at the very bottom of the providers list in /auth/providers It is very straightforward. Check the docs here: https://supabase.com/docs/guides/auth/custom-oauth-providers

Did you manage to set/edit providers or storage settings on the fly without needing a service restart ?

Turns out i'm on the same version, so idk thanks though !

I use cAdvisor and i log it into logflare. I have a graph showing it is continuously using about 250mb of ram. It's the same accross ~17 projects. My version is not up to date (last update was in feb) but since you've been using yours since 4 month i was wondering if you were on an older version that used less ram.

What pooler version are you using ? Mine is using x4 the ram of yours

Always wondered how yo do it, thanks !

Done

No, the architecture i have in mind is k3s managed (or individual vm in pve but the k3s way make more sense) So hetzner (server set up using ansible) -> debian (with or without pve) -> k3s managed by my next app -> supabase

Here was their answer, it's not allowed to talk about competing products/solutions in this sub... https://preview.redd.it/tppto8tlrvjg1.jpeg?width=1812&format=pjpg&auto=webp&s=31ce004b6d1d2f7a01435d9b248b453753ae1b08

Update: supabase edge functions are now supported, i can send you a link if you want to try it

To anyone wondering why this post got taken down, jsut know you are not allowed to talk about projects/solutions competing with supabase on this sub https://preview.redd.it/xbens1wxqvjg1.jpeg?width=1812&format=pjpg&auto=webp&s=967a2f3bd9237d3e89720c6dcb0bc05f3c08356a

Apparently we are not allowed to share projects competing with supabase on sub.... https://preview.redd.it/kfmx3v7pqvjg1.jpeg?width=1812&format=pjpg&auto=webp&s=18ac74f59999582052b35041c4d4b83fb7620f33

What a shame indeed

Will do :)

Sure thing, i'll send you a DM right now !

Thanks! I've actually already built most of these auth providers configuration, email templates, SMS setup, SQL editor, and more are all available in the dashboard UI of my solution. The only thing missing right now is JWT and secret key management from the UI, but that should be done by the end of the week. Feel free to send me a DM if you want to try it out!

Done !

Sure thing, i'll send it in your DMs

After looking it up the internet, you unfortunately cannot login with another instance than supabase's as the url is hardcoded in loveable .env though, if you selfhost it, you could edit the .env to use : VITE\_SUPABASE\_PROJECT\_ID="your-project-id" VITE\_SUPABASE\_PUBLISHABLE\_KEY="your-anon-key" VITE\_SUPABASE\_URL="https://your-custom-domain.com" So Lovable Cloud + Self-hosted Supabase doesn't work but Lovable Slef-hosted + Self-hosted Supabase can.

Yeah, the self-hosted experience can be pretty rough once you go beyond a basic setup. As the comment says lots of the friction comes from missing tooling, incomplete workflows, and the fact that the managed cloud version has quite a bit of extra glue and operational work behind the scenes that isn’t available out of the box. Part of what I’ve been building is specifically to smooth over that gap. I’ve added a number of features that aren’t present in the standard self-hosted version (and some that exist only in the managed cloud offering), especially around day-to-day usability and configuration. For example: * full project setup and configuration directly from the UI * auth management (providers, settings, templates, etc.) without touching the CLI * simplified deployment and project provisioning * additional management tooling that removes a lot of the manual configuration steps * a setup process that can be done entirely through the UI, no manual container tweaking required So the goal isn’t just “self-hosted Supabase cheaper”, but more like “self-hosted Supabase with the missing operational layer and quality-of-life tooling added back in while bieng managed”, while keeping the flexibility of dedicated instances. Totally agree though that raw self-hosting today still has rough edges, that’s actually one of the main problems I’m trying to solve

I'm European so server will be inn Europe too, but regarding external audits: it’s honestly still too early at this stage to engage auditors or start formal certifications. Right now the project is in a validation / early testing phase, and those kinds of processes come with significant costs and operational overhead.

I'll give it a look thanks for the tip

I have no clue to be honest but *i guess* if it uses mcp to connect to the db it's compatible

Thanks a lot for the thoughtful comment and honestly I completely agree with most of your reasoning, data hosting is serious business, and the pricing from existing DBaaS providers definitely reflects the operational complexity behind the scenes (security, uptime, backups, monitoring, etc.). You’re also right that technically anyone *can* self-host Supabase on a bare metal server and a lot of people do. The idea here isn’t to compete with “DIY self-hosting”, but more to sit in the middle ground between fully self-managed and enterprise-priced managed platforms. Basically: something affordable but still removing the operational burden. The goal is to make it viable through automation and density rather than massive margins per customer but you’re absolutely right that sustainability is the real challenge, which is why I’m testing demand early before investing too heavily. One thing I’ve been focusing on beyond just “cheap hosting” is quality-of-life features on top of the self-hosted stack. For example: * managing auth providers, settings etc.. directly from the UI * project configuration without touching the CLI * easier environment & project setup * UI improvements and additional management tooling to reduce the friction of self-hosting So the idea is less “cheap Supabase clone” and more “dedicated instance + managed experience + tooling improvements” while also being cheaper than supabse. Really appreciate the honest feedback though and you’re right to question the sustainability side, that’s exactly what I’m trying to validate right now :)

I understand, that's why backups, point in time recovery and high availability features are going to be a must have

I can send you a link to try it if you want, feel free to send me a dm :)

I'll send it in your DMs

Yeah, no bandwidth limit is what i consider to be a must have, I just hope no one will abuse it as I'd have to implement fair usage policies or rate limiting at some point if that happens. But for normal use it shouldn't be an issue

Ok, i get it now, thank you. I wasn't aware of the 100gb storage i thought it would be shared. In my case, the 30-40gb would be shared between the db and the file storage

Backups are not supported for now but it's the most important feature imo. Also the db being paused after a week is SOOO annoying, this won't be a thing in my solution ofc.

Yeah i'm not really fond of it either, I'm still planning on adding full support for it though

Send me a dm if you'd like to try it !

Good point, being hosted in Europe, this covers the data residency side of GDPR. As for the auth table, since each customer gets their own isolated Supabase instance, you have full control over your Postgres database including the auth schema. So you could implement data deletion requests, export user data, etc. directly through SQL if needed. I wouldn't touch the auth internals myself, but the fact that you have full database access means you're not locked out of anything, GDPR compliance for your app's users would be in your hands, which is arguably better than being on a platform where you can't access that data at all.

Thank you for your comment, I hesitated going that route too but fixed resource allocation allows my needs to be predictable, i might add several tiers later on as people seems to need more storage. Why didn't you release your solution btw ?

I agree, sadly automatic backups are not yet supported but It'll definitely be a must have along with other high availability features.

I considered AWS before going the VPS route, how much are you paying for your instance per month ?

I can sent you the link to try it in DMs if you want :)

30-40GB would be shared between db and file storage. Though, lots of people requested more storage so i might add storage tiers

I managed to make it work locally, i can spin up new instances from the UI and even manage stuff like auth providers without having to use the CLI. I can send you the link in dm if you're interested to try it.

Sent me a dm if you'd like to try it !