I've observed a peculiar issue when deploying solutions to production, particularly concerning the security settings of custom connectors. These connectors are automatically included in our solution when we add connection references to them. However, it seems that their security settings reset or change after deployment.
Example 1:
In our development environment, we have a custom connector using API Key authentication. The Parameter name is set to api-key. However, when this is imported to prod, it reverts to the default parameter name, which is Ocp-Apim-Subscription-Key. Once I manually correct this and update the connector in prod, it operates as expected.
I also checked the swagger, but they are identical (except for example 2) in both cases.
Example 2:
This situation is trickier. If we use the default Parameter name (Ocp-Apim-Subscription-Key) in the dev environment, after deployment, there's no visible change in the production settings. Despite this, the connector doesn't function correctly. The solution is to click 'Edit' under the security settings in prod and then immediately 'Update' the connector, without actually making any changes. Only then does it work.
In both scenarios, until these changes are made, we are met with errors such as "API Key not found" or something similar.
It's evident that some settings or parameters are being altered during the migration between environments, but pinpointing the exact cause has been challenging. Does anyone know of specific environment or deployment pipeline settings that could influence this? Or has anyone encountered a similar issue?
As a temporary fix, I've created a 'Custom Connector Tester' flow to verify functionality post-deployment, and I manually correct issues as they arise. However, this isn't a sustainable solution, especially since flows can be disrupted during the rectification process.
Any insights or solutions would be greatly appreciated!