testdrive: get rid of failed writing row to mz_aws_connections table

[nrainer-materialize]

This addresses https://github.com/MaterializeInc/database-issues/issues/8653.

Get rid of

[2m2024-10-11T06:09:53.863297Z[0m [31mERROR[0m [2mmz_adapter::catalog::builtin_table_updates[0m[2m:[0m failed writing row to mz_aws_connections table [3mid[0m[2m=[0ms759 [3me[0m[2m=[0minternal error: no AWS external ID prefix configured

[nrainer-materialize]

@benesch, should this work? (Spoiler: It does not yet.)

[benesch]

Hm, I didn't realize we already set the external ID prefix in the mzcompose test. Interesting! Can you link me to where you're seeing the error logs in Buildkite? Something I don't understand is going wrong here.

[nrainer-materialize]

Can you link me to where you're seeing the error logs in Buildkite? Something I don't understand is going wrong here.

Sure! An example is in "Testdrive 1", line 345.

[benesch]

Ahh! It's testdrive itself that's emitting the error. Not the materialized service (i.e., not in services.log). I pushed up a commit that should fix this.

[nrainer-materialize]

Thank you, @benesch!

I fixed the compilation issue but I fear that the log entry still occurs and we get a different error message:

2024-10-16T06:29:57.364989Z ERROR mz_adapter::catalog::builtin_table_updates:
failed writing row to mz_aws_connections table id=s759 e=internal error: no AWS connection role configured

when running bin/mzcompose --find testdrive down && bin/mzcompose --find testdrive run default --default-timeout 5s mz-sources.td

I will try to fix it.