Outbound REST Connector - Provisioning Retry Policy

Hello,

I have a question regarding outbound REST connector behavior in Memority.

Let’s take the following example:
A CREATE operation for Google Workspace performs two consecutive API calls:

1. User creation

2. License assignment

In this scenario, it is possible to receive an HTTP 429 (Too Many Requests) during the second call (license assignment).

According to the documentation, the provisioningRetryPolicy applies when the remote target application is unreachable (network failures).

I would like clarification on the following points:

1. Does provisioningRetryPolicy apply strictly to network-level errors (timeouts, DNS failures, connection refused), or can it also retry in case of HTTP application errors (e.g., 429, 403, 500)?

2. If no provisioningRetryPolicy is configured, does Memority automatically retry provisioning operations that fail due to application-level errors (HTTP 4xx/5xx), or are such errors considered final?

3. In the case of HTTP 429 (rate limiting), what would be the recommended approach?

   • Should I rely on a ProvisioningTaskDefinition to re-trigger provisioning?

   • Should retry logic be implemented directly inside the connector code?

   • Or is there a best practice mechanism within Memority to properly handle rate limit scenarios?

My concern is avoiding partially provisioned states (for example, a user successfully created but license assignment failing due to rate limits).

Thank you !