Rafael Felipe
09/03/2026, 5:59 PMNo catalog found in source discovery outputkapa.ai
09/03/2026, 5:59 PMdiscover operation ran but returned no catalog — the connector failed to enumerate any streams/tables. Based on the knowledge sources, here are the most common causes and fixes:
1. Permissions Issue The database user may lack the required privileges to read table metadata.
• Verify the credentials used in your source configuration have SELECT privileges on the target tables/schemas.
• For MySQL → PostgreSQL, confirm the MySQL user has the correct grants. [Forum: MySQL to Postgres]
2. Schema / Namespace Misconfiguration • If you are connecting to Postgres, try limiting the source to the public schema only in the connector settings. Multiple schemas have historically caused discovery to hang or return empty results. [Forum: Postgres schema detection failure]
• A known bug (fixed in Postgres connector v0.4.14) caused missing streams when multiple schemas were configured. Upgrading the connector resolved it. [Forum: Postgres schema detection failure]
3. Empty Table Set Discovered The CDK will throw a config error if zero tables are discovered:
> "Discovered zero tables." or "Unable to query any of the N discovered table(s)."
This can happen if:
• The configured schema/namespace is empty or the user has no access to any table.
• For SQL Server (MSSQL), system schemas are filtered out automatically; ensure your tables are in a non-system schema. [CheckOperation.kt]
4. Connector Version Bug • Source Mixpanel versions 3.6.0 and 3.6.1 had a regression causing a StopIteration error during discovery. Rolling back to v3.5.4 resolved it. [GitHub Issue #69098]
• Check if your connector version has known discovery bugs and upgrade/downgrade accordingly.
General Troubleshooting Steps 1. Check connector logs — open the connection in the Airbyte UI, go to the job logs, and look for the root cause error.
2. Open browser DevTools → Network tab and inspect the /discover API response for more details. [Forum: Postgres schema detection failure]
3. Verify database credentials have at least SELECT on all target tables.
4. Try a specific schema instead of discovering all schemas.
5. Upgrade your connector to the latest version to benefit from discovery bug fixes.
Could you share which source connector you're using and its version? That would help narrow down the exact cause. 👍 0 button 👎 0 button ℹ️ Tip: Mention @kapa.ai in the thread for follow-ups or DM the app for 1-to-1 help.Rafael Felipe
09/03/2026, 6:01 PMkapa.ai
09/03/2026, 6:01 PMsource-google-analytics-data-api):
1. Multiple Property IDs Causing Discovery Timeout (HTTP 504) This is a common reported issue. When 5–10 GA4 property IDs are configured in a single source on versions newer than 2.7.7, discovery generates a large number of streams and can timeout, causing the UI to show:
> An unknown error occurred. (HTTP 504)
even though the discovery pod itself completes successfully. [GitHub Issue #69745]
Fixes:
• If self-hosted on Kubernetes, increase the nginx ingress timeout by adding annotations to your Ingress resource: [GitHub Issue #71797]
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'
• Downgrade to connector version 2.7.7, which is confirmed to work with multiple property IDs. [GitHub Issue #69745]
• Split into multiple sources, one per property ID (reduces but may not fully eliminate timeouts). [GitHub Issue #69745]
2. Authentication Errors During Discovery If you are using OAuth credentials, discovery can fail with a 401 Unauthorized error if:
• Your refresh token or access token has expired or been revoked
• You are supplying custom OAuth credentials on Airbyte Cloud (which behaves differently from OSS)
Fix: Re-authenticate your source by editing it and refreshing your credentials. [GitHub Issue #62080]
3. Wrong Auth Schema Validation On newer connector versions, logs may show:
> JSON schema validation failed. $.auth_type: must be the constant value 'Client'
even when using a service account. This is a known validation mismatch on versions newer than 2.7.7 and is related to the multi-property discovery timeout issue. [GitHub Issue #69745]
Recommended next steps:
1. Check how many property IDs are configured — if more than one, that is likely the root cause.
2. Review your connector logs for 504, 401, or JSON schema validation errors.
3. If on Kubernetes, apply the nginx timeout fix above.
What connector version are you using, and how many property IDs do you have configured? 👍 0 button 👎 0 button ℹ️ Tip: Mention @kapa.ai in the thread for follow-ups or DM the app for 1-to-1 help.