Thanks a lot @dimitarvp . I have edited my post with the proper formatting for the error message.
What else is needed in terms of providing a configuration, other than the info I posted? Do you mean my mix file? Thanks a lot for trying to help!
Admittedly I never worked with Snowflake but have you tried connecting to it outside Elixir, with some dedicated tool? Similar to how people test connecting to their Postgres DB with the psql CLI tool.
yes, I just tested the actual connection with a Snowflake tool called snowsql that works just like psql and I can connect without any issue, using the same connection data I am using in my config file. I very much think the issue is in the ODBC setup on my local machine, but I am not being able to debug it as there are so many different pieces that need to work together.
Thanks a lot anyways for your time and your suggestions on how to improve my question.
not really, just that if the app is not in the list of apps that is started, the config is pruned. so it needs to be a name of an app that is started, not a random atom that you pass in the use Snowflex macro.
Hi! I’m afraid this won’t be too helpful, but let’s see:
I have looked extensively, but did not find any similar error
Yes, the errors are difficult to read/reason about (we had some connection errors that would not allow re-connecting, so eventually the machine would just die). We were only able to prove that it happened by attaching a connection listener to the connection and cross referencing the logs in the odbc driver. Once a connection died it would not come back up.
As far as your issue:
Unixodbc 2.3.12 (installed through Homebrew following the Snowflex readme).
This is the best hunch I have at the moment I say this because a lot has changed with the M1 and maybe that’s the issue? Maybe the version is not the right one? I’d check that first.
I’d open an issue on the Snowflex repo as well, as I bet the folks using it maybe have seen this issue before and know exactly how to handle it. The M1 has been around for a while now, I’m sure some folks that are using this have seen it before.
General thoughts on odbc
The ODBC route was difficult to use after a while. I’m not a fan of “here is a suggestion that changes all of your current workflow”. BUT, I was part of a project that dealt with this problem before and the result was moving away from odbc (last code commit in the erlang file was in 2010) and started using Snowflake’s api.