I'm trying to follow along with this docs page (https://supabase.com/docs/guides/platform/migrating-and-upgrading-projects#migrate-your-project) to export my database. When I run one of the db dump commands, like:

supabase db dump --db-url "postgresql://postgres:MYPASSWORD@db.krpyriciwjyezbzvkmjo.supabase.co:5432/postgres" -f schema.sql --debug

(replacing

MYPASSWORD

with the actual password for the postgres user) ...I get this:

Dumping schemas from remote database...
Error: Error running pg_dump on remote database: Cannot connect to the Docker daemon at unix:///var/run/docker.sock. Is the docker daemon running?

Two questions: 1. When the docs say to use the "database URL" is this what they mean? Am I supposed to use the full

postgresql:// (etc.)

connection string? 2. What am I doing wrong? Why can't I connect to the remote database? I have the latest versions of psql, pg_dump and the supabase CLI installed, and I've been able to export my database in the past using the old method in the docs of setting the postgres user as a SUPERUSER, then running a pg_dump command to dump the database.

I want to use types but the option to download them seems to be gone and only CLI is usable. I don't have docker installed and all I want from the CLI is the types. Is there anything I can do?

Hello, according to the documentation found on supabase, if you want to dispose of an instance you'd most likely do an

.removeChannel(someChannel)

What's then

myChannel.unsubscribe

used for then? in which cases is better to use

unsubscribe

instead of the other option?

Hello, I'm trying to find a way to implement RLS on my project. It is a event ticketing app where organizers can create new event and user can buy tickets. For the organizers I also have an app used for scanning tickets at the entrance, I'm doing a login using the eventID and a password set by the organizer through a rpc function. Once logged I need to get tickets from supabase (also with a realtime subscription) and update ticket status. Is there a way to implement RLS in some way or I necessarily need to login through supabase auth?

This article was posted and retweeted by the Supabase account and I'm curious on how it works. Does this prevent exposing the keys on the front end? https://twitter.com/doppler/status/1636377902244847616?s=20

Hello Supabase team, my question is how to setup in the API layer on Nextjs the server client respecting the RLS of the tables more context in this thread

I'm building an app in NextJS with a pretty big database structure up to now (~30 tables or so). These include users (with a high level app specific user_roles table to control in app access to features), teams (with a team_roles table), and different types of components that users can create based on their team etc. After doing some research, setting up a lot of complicated policies to deal with RLS and who gets to edit/delete/create what seems like it would be very complicated and time consuming to manage across the whole application. The other option is to simply enable RLS on all tables by default with no policies and have all supabase calls from within server components (I'm using the app directory) - however, with RLS enabled, and my server client setup to use the service role key - no data is being returned. Turning off RLS does allow the data to be returned (as one would expect). Any reason as to why this would be the case? I thought using the service role key bypasses all RLS policies. Thanks!

I have added a raise log line to my function

raise log 'var_where_clause = % ', var_where_clause;

When I go to Logs/Postgres there is 'No Results' - like no logging at all. Have I missed something? Or have I got it wrong?

Hey all! I've been fighting with trying to get local dev working on and off for months now and I can't seem to figure out what's going wrong. The main caveat I have is that I use podman rather than docker. I followed the steps here to run docker-compose with rootless podman (which got me past my initial blockers): https://fedoramagazine.org/use-docker-compose-with-podman-to-orchestrate-containers-on-fedora/ But whenever I run "supabase start" I get an error that the "Database is unhealthy...connection reset by peer" and subsequent containers don't start. Has anyone been able to get the CLI working with podman for local dev? Any help would be appreciated as I'm at a bit of a loss. Not sure if it's a configuration issue or maybe the health check retries expire before my container is ready. Thanks!

... solved it. I found the anon_key

How much memory and CPU does the Postgres database have available?

If anyone has gone through this before I would love some guidance on how you did it and your approach. Essentially I have a Google API key that I need to secure that's used for location search. What would be the best way of securing this? I know Supabase secrets exists but is this the way to go?

I created a bucket called

photos-bucket

and I wrote the following logic:

let { error: uploadError } = await supabase.storage
        .from('photos-bucket')
        .upload(filePath, file, { upsert: true });

But I get the error that it cannot read properties of undefined? It looks like

supabase.storage

is undefined but why is this the case?

I looked at the example,

js
const { error } = await supabase
  .from('countries')
  .update({ name: 'Australia' })
  .eq('id', 1)

and I'm confused. Is .update updating the data, or telling it what data to update? Same with .eq. What I want to do is change this (first image) to this (second image) with my code, but I don't understand how to. Also, sorry if I put this in the wrong tags, I don't really understand much.

Hi there! I'm currently in the process of building a tool with Langchain and I decided to run some benchmarks on vector stores. Interestingly, I found that there were significant differences in the results between Supabase (using pgvector extension) and Pinecone. Pinecone's results were much more accurate, but I'm not exactly sure why. I'm wondering if this could be due to Supabase not being as performant as Pinecone. As someone been benchmarking those tools or do anyone know the reasons behind this ? (to note, i've also created index using pg-vector in supabase and that help for speed for not for the quality of the results, the cosine distance seems not to been calculated the same way at tall)

I was able to deploy an edge function that calls on a random joke generator api and logs it in the console. For some reason my post requests after deploying are not being seen by the edge function and its not logged in logs or invocations.

Hi, I have a an "anomalies" table, I always fear that if something goes wrong in the frontend I might end up with a thousand requests to create an item in the table, how can I make sure this doesn't happen? I thought about limiting the allowed amount of entries per user with RLS, is this possible?

We use supabase to host Github OAuth of our website. We have a problem: 1. when the user login on our website for the first time, we get the github id and email of the user. 2. the user changed the main email on github 3. when the user login again on our website, the email is not updated to the new github email. How to solve this? Is there any code snippets? Thank you!

hi, in my app a user authenticates and then if its the first time they are taken to a profile page. if the user has already authenticated and entered profile info previously, I want the profile page to automatically route to the home/search page. If I do this with client-side code, its not great - a but slow with a few screen flickers. I tried to do with getserversideprops with some automatic routing like this: export async function getServerSideProps( ctx ) { const supabase = createServerSupabaseClient(ctx) const { req, res } = ctx; const { data: {user} } = await supabase.auth.getUser() if ({user}) { const { data, error } = await supabase .from('profiles') .select('created_at') .eq('id', user.id); if (data) { res.setHeader('Location', '/search'); res.statusCode = 307; res.end(); return { props: {} }; } return { props: {} }; } else { res.setHeader('Location', '/'); res.statusCode = 307; res.end(); return { props: {} }; } } user seems to be null during the auth flow - e.g. directly after authenticating using the Auth component (I think there is a similar issue if using withpageauth right after the Auth redirect). However, if I go back to the profile page once already logged in - the code works fine. Is there an issue in doing this when there is a redirect from the Auth component?

Im accessing my table from the client with:

supabase.from("designs").select("name");

As a test I used the default RLS policy of "Enable read access for all users" and it works. But I actually want to restrict access to logged in users. I made this policy but now the client gets an empty array instead of table data:

CREATE POLICY "Enable select for authenticated users only" ON "public"."designs"
AS PERMISSIVE FOR SELECT
TO authenticated
USING (true)

I used the email magic link authentication method. With the

useUser

hook I can see the user is authenticated (I can see the ID, email address, etc). Im brand new to Supabase. Any help debugging this would be much appreciated.

I signed up and can't perform any action. I confirmed my email address and still get unauthorized for every action I do. I can't even create a support ticket. Tried creating a new account with Github signup and I have the same issues, need this fixed asap.

When a user uploads a profile banner I want to take that image and split it into three images that are small, medium, and large. So when I make request for their banner my database just returns an object with the urls to the three different versions. Is this possible with supabase or would I have to use a 3rd party package to do this?

How supabase supports instance failures and how much involvement is needed?

Hi all. I'm trying to use Supabase with Qwik City and I want to create a server client on request, kind of like the Supabase auth helpers do with other full-stack frameworks. I'm using the code from the server-side rendering part of the docs to save the cookies under the website's domain so the server can access it. On request, I've been able to get these cookies and successfully call

setSession

, which validates everything and seems to give a proper response. However, when immediately calling

getSession

after setting the session it seems like the session was never set. Here is the code I'm using and the corresponding output:

const supabaseClient = createClient(import.meta.env.VITE_SUPABASE_URL, import.meta.env.VITE_SUPABASE_ANON_KEY)
const refreshToken = cookie.get("my-refresh-token");
const accessToken = cookie.get("my-access-token");

if (refreshToken && accessToken) {
  const res = await supabaseClient.auth.setSession({
  refresh_token: refreshToken.value,
    access_token: accessToken.value,
  });
console.log("Validated token when initializing client >>> " + JSON.stringify(res.data.session?.access_token))
console.log("Getting the new session right after >>> " + JSON.stringify(await supabaseClient.auth.getSession()))
}

Output:

Validated token when initializing client >>> "eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJhdWQiOiJhdXRoZW50aWNhdGVkIiwiZXhwIjoxNjc5MDU4NTI5LCJzdWIiOiJiNjIyMTY2Ny1hZGQ2LTQ2YWEtOWM1Ny0xOTNlNjMyNmIzMWUiLCJlbWFpbCI6ImZvb2Zvb0BiYXJiYXIuY29tIiwicGhvbmUiOiIiLCJhcHBfbWV0YWRhdGEiOnsicHJvdmlkZXIiOiJlbWFpbCIsInByb3ZpZGVycyI6WyJlbWFpbCJdfSwidXNlcl9tZXRhZGF0YSI6e30sInJvbGUiOiJhdXRoZW50aWNhdGVkIiwiYWFsIjoiYWFsMSIsImFtciI6W3sibWV0aG9kIjoicGFzc3dvcmQiLCJ0aW1lc3RhbXAiOjE2Nzg5NjI5MTh9XSwic2Vzc2lvbl9pZCI6IjhhMWIzYjNjLWYxMmMtNDcwYy05MDg0LWRhZDZkYjlkMGE2YiJ9.ehZMXCMa_FVmY1kXdLkNlkorbVw1ghKJpEWWxEnPYzQ"
Getting the new session right after >>> {"data":{"session":null},"error":null}

Thanks for your help! I'm very confused by this

Hello! I noticed that the only thing the signout method does is to remove the session token from the browser. So if I sign out from the profile page, I am still left on the profile page, until I reload or navigate elsewhere. Clicking signout "should" bring you back to the login page. If I use an event handler and route to "/". That would bring you back to the login page because I have middleware that will handle that. However, I would need to use a setTimeout since there is a bit of a delay before the session token is fully removed. Is there a better way to handle that? One would think that the supabase function would take care of the full sign out process. The NextJS tutorial says to use it on a button like this:

<button className="button block" onClick={() => supabase.auth.signOut()}>

However the docs for sign out show that we should use await:

const { error } = await supabase.auth.signOut()

supposedly it is supposed to trigger a sign out event but nothing visibly happens. Thank you!

createServerSupabaseClient is using NEXT_PUBLIC_SUPABASE_ANON_KEY by design (on nextjs helpers) is there any way to bypass this?

hi! I'm having this error: running db migrations: error executing migrations/20230116124310_alter_phone_type.up.sql, sql: -- alter phone field column type to accomodate for soft deletion alter table auth.users alter column phone type text, alter column phone_change type text; : ERROR: cannot alter type of a column used by a view or rule (SQLSTATE 0A000) and I cant login to my app, how could I fix this? thanks!

Is it possible to get data from authentication using the SQL on the supabase platform?

I've checked this resource https://supabase.com/docs/guides/auth/row-level-security#policies-with-joins and tried to recreate it, but it always returns zero values. My tables in public: - profiles (which equals auth.user.id, so auth.uid() should match this id) - teams - team_users (join table for the two above with user_id and team_id) My policy on the

teams

table should allow authenticated users to retrieve teams they belong to: My USING expression is the following:

pgsql
id IN (
  SELECT user_id
    FROM public.team_users tu
    WHERE tu.user_id = auth.uid()
)

What is wrong with this expression? Why does it return zero results while my user clearly belongs to two teams?

I am creating a local development env with a fair amount of data and I'd like to export it to a seed.sql file. The docs say to use

supbase db export

but that command doesn't seem to exist. Any help would be greatly appreciated!