When configuring Service Engine (SE) Groups under Tenant Mode, inconsistencies are observed while overriding the management network, depending on the tenant_vrf configuration:
Although the error is correctly returned, the behavior is inconsistent. If the user cancels the SE Group configuration and edits it again, the updated management network settings are visible, indicating that the change was partially applied.
If the user refreshes the page or cancels the operation and reopens the SE Group configuration, the updated management network is visible.
VMware Avi Load Balancer Versions 22.1.x, 30.2.x till 30.2.6, 31.1.x and 31.2.1
The way the user interface handles the backend API requests during the save process is the cause for these issues
A PUT request to /api/serviceenginegroup/<uuid>: This request succeeds and updates the mgmt_network_ref field in the SE Group object.
A PUT request to /api/vrfcontext/<uuid>This request fails with HTTP status code 400, resulting in the displayed error. Despite the error, the SE Group object has already been updated, which explains why the new management network is visible upon re-editing.
The PUT request to /api/serviceenginegroup/<uuid> succeeds, updating the management network. However, the UI does not properly handle the completion of the request flow, causing the page to remain stuck even though the configuration change has been applied in the backend.
These inconsistencies in SE Group management network override behavior under Tenant Mode have been fixed in the following releases: