I googled it but only saw some unanswered question - mainly in outbound “message boards” w.r.t. to messages like this:

Metaflow 2.12.19 executing CDCNSLTest for user:me_me_me
Validating your flow...
    The graph looks good!
Running pylint...
    Pylint is happy!
2024-09-11 10:24:37.000 Bootstrapping virtual environment(s) ...
2024-09-11 10:24:37.243 Virtual environment(s) bootstrapped!
    Metaflow service error:
    Metadata request (/flows/SomeTest) failed (code 500): {"message": "Internal server error"}

I am pretty sure, that in my particular use case this error is due to the fact that, a devops (TBC) upgraded the meta data postgresql db in our Terraformed Metaflow AWS infra. I was told that we cannot use the older version anymore due to costs/AWS support. Is there any quick fix one could use? Maybe metaflow does not support later version of postgresql, which would surprise me … Any quick (sorry) help would be appreciated please! Thanks! PS: It appears to have to do with the DDL (?) Python code used to create the metaflow postgresql database - it does not play ball with the postgresql 15 due to some SSL related changes?

are you using the latest version of the metadata service? that supports ssl

I will ask thanks.

the default option group for RDS starting from Postgres engine 15 onward sets

force_ssl

to true. Metaflow service introduced ssl support for the connection string in

2.4.6

onward, but the default for it is still disabled. deploying the services with the correct version should therefore also have the environment variable set for example as

MF_METADATA_DB_SSL_MODE=prefer

@square-wire-39606 thanks devops bumped everything and things look good again so far. thanks!