Snowflake ingestion in error

[FabriceArchiData]

Hello,

We have just installed OpenMetaData on Docker, we configured the Snowflake connection, it's OK.
When we'll ingest the metadata (on only one DB, one schema), we have this error :

_

`[2026-07-23 14:58:09] WARNING - Retrying (Retry(total=0, connect=None, read=None, redirect=None, status=None)) after connection broken by 'ReadTimeoutError("HTTPSConnectionPool(host='yj12ezsfcb1stg.blob.core.windows.net', port=443): Read timed out. (read timeout=7)")': /results/01c5e6aa-030a-043c-0000-07c1d2cab24a_0/main/data_0_2_0_2?sv=2026-02-06&se=2026-07-23T18%3A58%3A02Z&skoid=76d3b8de-f998-4457-9806-888624166ce1&sktid=8b8389f7-0e38-4006-b754-f756e0ac6052&skt=2026-07-23T12%3A25%3A32Z&ske=2026-07-30T12%3A24%3A32Z&sks=b&skv=2026-02-06&sr=b&sp=r&sig=&rsce=gzip source=snowflake.connector.vendored.urllib3.connectionpool loc=connectionpool.py:869
[2026-07-23 14:58:16] ERROR - Failed to fetch the large result set batch data_0_0_1_2 for the 1 th time, backing off for 1s for the reason: 'HTTPSConnectionPool(host='yj12ezsfcb1stg.blob.core.windows.net', port=443): Max retries exceeded with url: /results/01c5e6aa-030a-043c-0000-07c1d2cab24a_0/main/data_0_0_1_2?sv=2026-02-06&se=2026-07-23T18%3A58%3A02Z&skoid=76d3b8de-f998-4457-9806-888624166ce1&sktid=8b8389f7-0e38-4006-b754-f756e0ac6052&skt=2026-07-23T12%3A25%3A32Z&ske=2026-07-30T12%3A24%3A32Z&sks=b&skv=2026-02-06&sr=b&sp=r&sig==gzip (Caused by ReadTimeoutError("HTTPSConnectionPool(host='yj12ezsfcb1stg.blob.core.windows.net', port=443): Read timed out. (read timeout=7)"))' source=snowflake.connector.result_batch loc=result_batch.py:431
ConnectionError: HTTPSConnectionPool(host='yj12ezsfcb1stg.blob.core.windows.net', port=443): Max retries exceeded with url: /results/01c5e6aa-030a-043c-0000-07c1d2cab24a_0/main/data_0_0_1_2?sv=2026-02-06&se=2026-07-23T18%3A58%3A02Z&skoid=76d3b8de-f998-4457-9806-888624166ce1&sktid=8b8389f7-0e38-4006-b754-f756e0ac6052&skt=2026-07-23T12%3A25%3A32Z&ske=2026-07-30T12%3A24%3A32Z&sks=b&skv=2026-02-06&sr=b&sp=r&sig==gzip (Caused by ReadTimeoutError("HTTPSConnectionPool(host='yj12ezsfcb1stg.blob.core.windows.net', port=443): Read timed out. (read timeout=7)"))
MaxRetryError: HTTPSConnectionPool(host='yj12ezsfcb1stg.blob.core.windows.net', port=443): Max retries exceeded with url: /results/01c5e6aa-030a-043c-0000-07c1d2cab24a_0/main/data_0_0_1_2?sv=2026-02-06&se=2026-07-23T18%3A58%3A02Z&skoid=76d3b8de-f998-4457-9806-888624166ce1&sktid=8b8389f7-0e38-4006-b754-f756e0ac6052&skt=2026-07-23T12%3A25%3A32Z&ske=2026-07-30T12%3A24%3A32Z&sks=b&skv=2026-02-06&sr=b&sp=r&sig=*&rsce=gzip (Caused by ReadTimeoutError("HTTPSConnectionPool(host='yj12ezsfcb1stg.blob.core.windows.net', port=443): Read timed out. (read timeout=7)"))
ReadTimeoutError: HTTPSConnectionPool(host='yj12ezsfcb1stg.blob.core.windows.net', port=443): Read timed out. (read timeout=7)
TimeoutError: select timed out
Fichier "/home/airflow/.local/lib/python3.10/site-packages/snowflake/connector/vendored/urllib3/connectionpool.py", ligne 464 dans _make_request

Fichier "/home/airflow/.local/lib/python3.10/site-packages/snowflake/connector/vendored/urllib3/connectionpool.py", ligne 1106 dans _validate_conn

Fichier "/home/airflow/.local/lib/python3.10/site-packages/snowflake/connector/vendored/urllib3/connection.py", ligne 805 dans connect

Fichier "/home/airflow/.local/lib/python3.10/site-packages/snowflake/connector/vendored/urllib3/connection.py", ligne 984 dans _ssl_wrap_socket_and_match_hostname

Fichier "/home/airflow/.local/lib/python3.10/site-packages/snowflake/connector/ssl_wrap_socket.py", ligne 194 dans ssl_wrap_socket_with_cert_revocation_checks

Fichier "/home/airflow/.local/lib/python3.10/site-packages/snowflake/connector/vendored/urllib3/util/ssl_.py", ligne 433 dans ssl_wrap_socket

Fichier "/home/airflow/.local/lib/python3.10/site-packages/snowflake/connector/vendored/urllib3/util/ssl_.py", ligne 477 dans _ssl_wrap_socket_impl

Fichier "/home/airflow/.local/lib/python3.10/site-packages/snowflake/connector/vendored/urllib3/contrib/pyopenssl.py", ligne 522 dans wrap_socket

WantReadError:
Fichier "/home/airflow/.local/lib/python3.10/site-packages/snowflake/connector/vendored/urllib3/contrib/pyopenssl.py", ligne 519 dans wrap_socket

Fichier "/home/airflow/.local/lib/python3.10/site-packages/OpenSSL/SSL.py", ligne 2468 dans do_handshake

Fichier "/home/airflow/.local/lib/python3.10/site-packages/OpenSSL/SSL.py", ligne 2040 dans _raise_ssl_error`

_

We check the network, between OpenMetaData server and the blob the curl command is OK.

Do you have an idea how to resolve this error ?
Where can we alter the read timeout ? (I try on OpenMetaData interface, I added the parameter readtimout to 360 in Snowflake parameters) but nothing.

Best Regards,
Fabrice

[yhay81]

This is not a Snowflake login/query timeout. The query succeeds; the Python connector then fails while downloading a large result batch from Snowflake's Azure stage. That is why “Test Connection” can pass.

First, please edit the post and redact everything after .blob.core.windows.net/.... Later traceback lines still contain the signed Azure sig= query parameter. It is short-lived/read-only, but should not be public.

Snowflake documents this exact split: results over 100 KB require full outbound access and certificate passthrough to every STAGE endpoint: Fetching large query result sets failures.

I would troubleshoot it in this order:

1. In Snowflake run SELECT SYSTEM$ALLOWLIST(); (or SYSTEM$ALLOWLIST_PRIVATELINK()) and allow every returned STAGE host on TCP 443. Do not assume one current *.blob.core.windows.net hostname is permanent.
2. Run the test from the Airflow/ingestion container, because the traceback is under /home/airflow; a curl from the Docker host or OpenMetadata server container does not test the same DNS, proxy, CA bundle, or egress policy. Snowflake recommends SnowCD with that allowlist.
3. Check any proxy/TLS-inspection device. A successful HEAD/curl to the hostname does not prove that the connector can complete a large signed-object GET with certificate passthrough.

For the timeout, the parameter is spelled socket_timeout, and it belongs under OpenMetadata Advanced Configuration → Connection Arguments, not Connection Options:

connectionArguments:
  socket_timeout: 360
  network_timeout: 600

socket_timeout controls socket-level reads/connects; network_timeout controls retry time for non-login requests: Snowflake connector timeout docs. OpenMetadata passes connectionArguments to the driver and currently defaults network_timeout to 600: connection implementation.

So readtimout has two problems: it is misspelled and is not a Snowflake connector argument. Raising socket_timeout may confirm a merely slow link, but a repeated TLS/read failure at the stage host should be fixed in the container's network/allowlist path first.