upstream connect error or disconnect/reset before …d and the latest reset reason: connection timeout
Suddenly, my app is behaving in a very odd way. The browser console shows a lot of errors such as: upstream connect error or disconnect/reset before …d and the latest reset reason: connection timeout"
25 Replies
Are you looking at a specific page in the console?
Something that works with realtime or anything else or is this just general across the application?
The whole system became very very slow
Do you happen to have the error log from console you can share?
I thought it could be due to heavy usage in my free plan but I upgraded to Pro. Maybe it takes a while?
I'm using it now and I can't see errors anymore, but things are very very very slow
have you hard a recent burst in users or activity?
my suggestion might be to go into the dashboard -> reports and just flick through the different reports starting with
API Gateway/ PostgRESTNo, it's just me.
When I see the reports, everything seems fine, which is very awkward 😐
Can someone look at my case, please? Or let me know if there is another communication channel for paying customers.
This is a user helping user forum. Support is the avenue if users have nothing for you.
Upgrading to Pro without upgrading your CPU from nano will change nothing performance wise.
Thanks for the clarification, @garyaustin.
I think Nano is enough for me
Pro includes the next level for free.
But it is not automatic as it requires a reboot.
Do I have to manually do the reboot? Or it happens automatically at some point?
You upgrade I think. I don't have Pro. But look at the CPU/memory section and see what options are there.
I'm often getting time outs. Example:
[Failed to load resource: The request timed out.](https://[my instance here].supabase.co/rest/v1/messages?select=*&campaign_id=in.(e48708f4-dda7-4cca-b03a-58cbef194833%2C61c6ff9e-eb85-418a-886b-2e4153f1b3ac%2C4ede4974-f02b-4be6-b686-6c99c72de3a9)&conversation_uuid=is.null&read_at=is.null&sender_id=neq.e53d055d-6bbf-4747-8950-db9ff989554d)
Check the query reports under database and see if you have queries/RLS that needs tuning.
Or [Error] Failed to load resource: the server responded with a status of 503 (Service Unavailable) (messages, line 0)
Hard to know that is pretty generic.
It's probably related to that, right 😐
The timeouts and stuff
specific query timeouts could be especially if the same ones.
Look in the query reports for slow queries.
RLS on large tables if not following the tuning guide can be a big hit.
RLS on large tables if not following the tuning guide can be a big hit.
Do you think in the Pro version, someone from Supabase could join a call with me to discuss these things?
They don't do that and rarely help with general performance if not a bug on their side.
🙁 that's unfortunate
Now the system just stopped working, as all queries are getting timed out
I upgraded to Micro
Ok, that solved all my issues apparently 🤯
You should look at the query reports.
Do you have large 100k> row tables?
If so you need to read the RLS tuning docs.
No table has more than 300 rows. But I did followed the tuning suggestions
Ok, there is one wiht 2k rows
That should not be enough for RLS to slow your queries.
Thanks for all the help, @garyaustin !