Hi all, this is intended as a helper to others/bug report, although I haven't figured out where in the stack the bug should be submitted.
Basically, I am connecting to a Snowflake "database" using odbc, and it works fine, unless the query is very large, which is apparently a feature: https://community.snowflake.com/s/article/CURLerror-SSL-peer-certificate-or-SSH-remote-key-was-not-OK-during. That link just says that Snowflake has this intended behavior that for very large queries, it bypasses the server and goes right to S3.
However, when working from my work laptop, this direct connection to S3 is not allowed by some enterprise software called Netskope. I get the following error:
Error in result_fetch(res@ptr, n) : nanodbc/nanodbc.cpp:2593: HY000: [Snowflake][Snowflake] (25)
Result download worker error: Worker error: [Snowflake][Snowflake] (4)
REST request for URL [actual URL removed for privacy] failed: CURLerror (curl_easy_perform() failed) - code=60 msg='SSL peer certificate or SSH remote key was not OK'.
I have been working with my awesome IT team to resolve this, and it looks like the "rsession.exe" process has to be explicitly allowed by Netskope, and supplied with allowed URLs. This is an acceptable fix (if certain assumptions prove true) but I wonder if this is something that could be handled in future odbc releases or not (i.e. with a parameter for credentials, etc). If not, I hope this will at least help the next person who googles the error.