[EPIC] Improve database sources & destinations error messages

[sherifnada]

Tell us about the problem you're trying to solve

When users are setting up a database connector and fail (due to misconfiguration) they are often provided a non-descriptive error message like the following:

This causes drop off in activation when users are trying out Airbyte. We'd like to make these error messages more descriptive as they can greatly increase the usage of Airbyte.

Describe the solution you’d like

I would like these error messages to be much more descriptive in the following connectors (ordered by priority):
Sources:

• Postgres
• MySQL
• MSSQL
• Snowflake
• Mongo
• File source
• Oracle

Destinations:

• Redshift
• GCS
• MySQL
• Postgres
• Mongo

I imagine the best way to achieve this:

1. Induce as many configuration failures as we can think of in test cases
2. Identify which error messages returned by the database correspond to which errors
3. Return helpful error messages from the check method to help the user understand the action they can take to fix the problem
4. Write test cases to ensure that these error messages are always helpful to the user

Some error cases we can run into (this is not comprehensive, just the ones I can think of right now):

• Wrong password (but correct everything else)
• Wrong username (but correct everything else)
• Wrong host (but correct everything else)
• Wrong port (but correct everything else)
• Wrong database name (but correct everything else)
• No permissions on the selected schema/DB/table
• Any other incorrect / incompatible settings

[tuliren]

More examples can be found in this slack thread.

[grishick]

The fix should probably use the new AirbyteTraceMessage that is being added to the protocol: #12458

[Phlair]

@grishick I'm currently working on this to create AirbyteTraceMessage functionality for Java connectors, after which it can be used to create the improved errors as per this issue.

Had an idea for some incremental value here by doing this: https://github.com/airbytehq/airbyte-internal-issues/issues/601
We looked at it in backlog grooming and think that jdbc ticket likely fits better in dbs & dws connector team, just fyi.

[mkhokh-33]

@Phlair, @grishick, @alexandr-shegeda

We had a look on trace messages documentation and we don't have clear understanding where and how this trace messages functionality should be applied in scope of this task.

We are working in scope of this epic with error messages marked on screenshot as section 2

So we are trying to map returned errors into user friendly error messages.
From documentation we see screenshots for log section with preview which we currently do not have

Could you please provide understanding what should be done in scope of trace messages functionality for this epic and UI mock ups how it should look ?

Thus we don't have clear understanding of trace messages role in scope of this ticket we propose to finish this epic without trace messages implementation and create and estimate new task for it, cause this changes might include some UI tasks as well.

Thanks

[alexandr-shegeda]

PR is already available for review here

[grishick]

Are this ticker and PR still active?

[mkhokh-33]

yes, will merge last updates form master