The programmability of upcoming transport optical networks is handled by a centralized Transport SDN (T-SDN) controller. The T-SDN controller dynamically computes and selects the resources to automatically accommodate optical connection requests by configuring the involved network elements and devices (i.e., optical switches and transceivers). Additionally, the T-SDN controller should provide management functions to ensure high reliability and correct service status in terms of QoS, QoT, failure detection, etc. during the whole connection duration. These management functions are typically addressed by the operation, administration and maintenance (OAM) Handler. Thereby, the OAM Handler interacts with the monitoring system to gather notifications about connection failures or degradations and trigger the required actions to preserve / recover the services. This work first overviews the designed and implemented architecture integrating required OAM functions into a deployed T-SDN controller for flexi-grid optical networks. Next, two restoration routing approaches are presented, which are compared with respect to the blocked bandwidth ratio (BBR), restorability and average number of used sub-transceivers. The experimental performance is made varying both the traffic load and the generation of signal degraded events of existing connections, forcing their automatic restoration.