Persistent connection timeouts after implementing reconnection logic
edge functionsHi Supabase team,
I'm experiencing persistent connection timeout errors with my Supabase project, and I'd appreciate guidance on what might be causing this.
Error Details:
upstream connect error or disconnect/reset before headers.
retried and the latest reset reason: connection timeout
Context:
This error occurs on GET requests to /rest/v1/raffles
Response times: 10-25 seconds before timeout
Using Next.js 16.1.6 with App Router (API routes)
Requests eventually return 200 but with the error message
This is intermittent - sometimes requests succeed quickly (<300ms), other times timeout
What I've Implemented:
Extended timeout to 25 seconds (fetchWithTimeout)
Connection keep-alive headers
Retry logic with exponential backoff
Lean projections (only 11 fields selected)
Limit of 24 records
Using cache: 'no-store' for fresh data
Current Request Configuration:
const res = await fetchWithTimeout(url.toString(), {
headers: {
apikey: anonKey,
Authorization:
Accept: 'application/json',
Connection: 'keep-alive',
},
cache: 'no-store',
timeout: 25000
})
Questions:
Is there a known issue with PostgREST connection pooling that could cause these timeouts?
Are there recommended timeout values for production apps?
Should I be using a different approach for fetching data (e.g., Supabase JS client vs direct REST calls)?
Could this be related to the ongoing backup infrastructure maintenance (Jan 26 - Feb 2)?
I'm experiencing persistent connection timeout errors with my Supabase project, and I'd appreciate guidance on what might be causing this.
Error Details:
upstream connect error or disconnect/reset before headers.
retried and the latest reset reason: connection timeout
Context:
This error occurs on GET requests to /rest/v1/raffles
Response times: 10-25 seconds before timeout
Using Next.js 16.1.6 with App Router (API routes)
Requests eventually return 200 but with the error message
This is intermittent - sometimes requests succeed quickly (<300ms), other times timeout
What I've Implemented:
Extended timeout to 25 seconds (fetchWithTimeout)
Connection keep-alive headers
Retry logic with exponential backoff
Lean projections (only 11 fields selected)
Limit of 24 records
Using cache: 'no-store' for fresh data
Current Request Configuration:
const res = await fetchWithTimeout(url.toString(), {
headers: {
apikey: anonKey,
Authorization:
Bearer ${anonKey}Accept: 'application/json',
Connection: 'keep-alive',
},
cache: 'no-store',
timeout: 25000
})
Questions:
Is there a known issue with PostgREST connection pooling that could cause these timeouts?
Are there recommended timeout values for production apps?
Should I be using a different approach for fetching data (e.g., Supabase JS client vs direct REST calls)?
Could this be related to the ongoing backup infrastructure maintenance (Jan 26 - Feb 2)?
