Nahid Oulmi
07/09/2026, 1:25 PMdiscover_schema step using the airbyte/source-sftp-bulk connector on Airbyte 2.0.1. Can you help ?kapa.ai
07/09/2026, 1:26 PMdiscover_schema typically means the schema discovery process is taking longer than your proxy/ingress timeout allows. Here are the relevant fixes:
1. Increase Airbyte Server Timeouts (Helm/Kubernetes) If you're running Airbyte on Kubernetes via Helm, update your `values.yaml`:
server:
httpIdleTimeout: 20m
extraEnv:
- name: READ_TIMEOUT
value: 30m
[Schema Discovery Timeouts]
2. Increase Nginx Ingress Timeouts If Airbyte sits behind an nginx ingress controller, the default timeout is 60 seconds, which is often too short for schema discovery. Add these annotations to your Ingress resource:
apiVersion: networking.k8s.io/v1
kind: Ingress
metadata:
name: airbyte-ingress
namespace: airbyte
annotations:
nginx.ingress.kubernetes.io/proxy-read-timeout: '600'
nginx.ingress.kubernetes.io/proxy-send-timeout: '600'
[GA4 Discovery Timeout Fix]
3. Docker Deployment: Increase Proxy Timeout If you're on a Docker-based deployment, you can set the BASIC_AUTH_PROXY_TIMEOUT environment variable:
BASIC_AUTH_PROXY_TIMEOUT=600
[Nginx 504 Forum Post]
4. Known SFTP-Specific Issue: SSH Keepalive There is a known issue with source-sftp-bulk where large file transfers (e.g., ~375 MB) can cause a paramiko.buffered_pipe.PipeTimeout because intermediate network devices (firewalls, NAT, load balancers) drop idle SSH connections. A fix has been submitted that adds SSH keepalive packets every 60 seconds to prevent this. [SFTP Keepalive Fix]
In the meantime, as a workaround:
⢠Check if your SFTP server or network has aggressive idle/session timeout settings and adjust them server-side.
⢠Verify whether smaller files transfer successfully through the same pipeline.
⢠Ensure there are no firewalls or proxies between Airbyte and your SFTP server that might drop long-running connections. [SFTP Bulk Issue]
Summary If the issue persists after applying these changes, check your browser's network tab for the exact API response on the discover_schema call, and review Airbyte server logs for more detail. š 0 button š 0 button ā¹ļø Tip: Mention @kapa.ai in the thread for follow-ups or DM the app for 1-to-1 help., with interactive elementsNahid Oulmi
07/09/2026, 1:36 PMdiscover_schema instead ?kapa.ai
07/09/2026, 1:36 PMsource-sftp-bulk specifically.
However, based on what the knowledge sources do describe, there is a relevant alternative:
Use the "Schemaless" Schema Option For file-based sources (which source-sftp-bulk is), Airbyte offers a schemaless schema option that avoids a time-consuming discover phase entirely. Instead of inferring the schema from your files, it uses a hardcoded schema of {"data": object}, wrapping each record under a single data key.
> "This option avoids a time-consuming or inaccurate discover phase and guarantees that everything ends up in your destination, at the expense of Airbyte being able to structure the data into different columns."
[Schemaless Sources]
Trade-offs to be aware of:
⢠ā No typed columns in the destination ā all data lands in a single data field
⢠ā Column selection and schema evolution features are not available
⢠ā
Discovery is fast and reliable
⢠ā
All records are guaranteed to reach the destination
Whether this option is exposed in the source-sftp-bulk connector's configuration UI is something you'd need to verify directly in the connector settings. If it is available, it would be the closest thing to "skipping" schema discovery. š 0 button š 0 button ā¹ļø Tip: Mention @kapa.ai in the thread for follow-ups or DM the app for 1-to-1 help.