Intermittent RLS Violation on `storage.object.sign_upload_url` Across Different Storage API Instance
storageWe are experiencing an intermittent
After comparing the raw logs from both a failed and a successful request, we identified that the only meaningful difference is the internal host handling the request:
- Failed request:
- Successful request:
This strongly suggests that certain storage API instances have a stale or inconsistent state regarding RLS policy evaluation, causing requests routed to those instances to fail while identical requests routed elsewhere succeed.
Project ref:
Region:
Operation:
Client:
Failing request trace ID:
Successful request trace ID:
new row violates row-level security policy42501createSignedUploadUrlsupabase-js-node/2.95.3After comparing the raw logs from both a failed and a successful request, we identified that the only meaningful difference is the internal host handling the request:
- Failed request:
ip-10-114-43-181.us-west-2.compute.internal- Successful request:
ip-10-114-4-238.us-west-2.compute.internalThis strongly suggests that certain storage API instances have a stale or inconsistent state regarding RLS policy evaluation, causing requests routed to those instances to fail while identical requests routed elsewhere succeed.
Project ref:
aqqcaqvqrpxtywfczsnrRegion:
us-west-2Operation:
storage.object.sign_upload_urlClient:
supabase-js-node/2.95.3Failing request trace ID:
9d34136b649ec02a-GRUSuccessful request trace ID:
9d343a1a13f81aad-GRU
