diff --git a/internal/pkg/onuadaptercore/omci_ani_config.go b/internal/pkg/onuadaptercore/omci_ani_config.go
index e50bd58..9cb7e4c 100644
--- a/internal/pkg/onuadaptercore/omci_ani_config.go
+++ b/internal/pkg/onuadaptercore/omci_ani_config.go
@@ -1170,7 +1170,33 @@
 		return
 	}
 }
-
+func (oFsm *uniPonAniConfigFsm) handleOmciAniConfigSetFailResponseMessage(ctx context.Context, msgObj *omci.SetResponse) {
+	//If TCONT fails, then we need to revert the allocated TCONT in DB.
+	//Because FSMs are running sequentially, we don't expect the same TCONT hit by another tech-profile FSM while this FSM is running.
+	oFsm.mutexPLastTxMeInstance.RLock()
+	defer oFsm.mutexPLastTxMeInstance.RUnlock()
+	if oFsm.pLastTxMeInstance != nil && msgObj.EntityClass == oFsm.pLastTxMeInstance.GetClassID() &&
+		msgObj.EntityInstance == oFsm.pLastTxMeInstance.GetEntityID() {
+		switch oFsm.pLastTxMeInstance.GetName() {
+		case "TCont":
+			//If this is for TCONT creation(requestEventOffset=0) and this is the first allocation of TCONT(so noone else is using the same TCONT)
+			//We should revert DB
+			if oFsm.requestEventOffset == 0 && !oFsm.tcontSetBefore && oFsm.pUniTechProf.mapPonAniConfig[oFsm.uniTpKey] != nil {
+				logger.Debugw(ctx, "UniPonAniConfigFsm TCONT creation failed on device. Freeing alloc id", log.Fields{"device-id": oFsm.deviceID,
+					"alloc-id": oFsm.pUniTechProf.mapPonAniConfig[oFsm.uniTpKey].tcontParams.allocID, "uni-tp": oFsm.uniTpKey})
+				if pDevEntry := oFsm.pDeviceHandler.getOnuDeviceEntry(ctx, false); pDevEntry != nil {
+					pDevEntry.freeTcont(ctx, oFsm.pUniTechProf.mapPonAniConfig[oFsm.uniTpKey].tcontParams.allocID)
+				} else {
+					logger.Warnw(ctx, "Unable to get device entry! couldn't free tcont",
+						log.Fields{"ME name": oFsm.pLastTxMeInstance.GetName(), "device-id": oFsm.deviceID})
+				}
+			}
+		default:
+			logger.Warnw(ctx, "Unsupported ME name received with error!",
+				log.Fields{"ME name": oFsm.pLastTxMeInstance.GetName(), "result": msgObj.Result, "device-id": oFsm.deviceID})
+		}
+	}
+}
 func (oFsm *uniPonAniConfigFsm) handleOmciAniConfigSetResponseMessage(ctx context.Context, msg OmciMessage) {
 	msgLayer := (*msg.OmciPacket).Layer(omci.LayerTypeSetResponse)
 	if msgLayer == nil {
@@ -1190,10 +1216,7 @@
 			log.Fields{"device-id": oFsm.deviceID, "Error": msgObj.Result})
 		// possibly force FSM into abort or ignore some errors for some messages? store error for mgmt display?
 
-		//FIXME: If setting TCONT fails we need to revert the DB back. Because of the concurency,
-		//doing it here may cause a data inconsistency. To fix this problem we need to think on running
-		//the FSMs of different UNIs sequentially instead of running them concurrently.
-
+		oFsm.handleOmciAniConfigSetFailResponseMessage(ctx, msgObj)
 		return
 	}
 	oFsm.mutexPLastTxMeInstance.RLock()