Controlling switch mechanism for detecting fibre channel over Ethernet data forwarder failure
Techniques are provided for detecting failure of switching elements of a distributed switch configured to forward Fiber Channel over Ethernet (FCoE) frames. Periodic discovery advertisement messages, which include a fabric name fieldm, are transmitted between a controlling Fiber Channel Forwarder (cFCF) and a FCoE data forwarder (FDF). The cFCF determines a FDF has failed out of the switch fabric based on a reserved value in the fabric name field of a received discovery advertisement message, such as a zero or null value, and de-instantiates virtual links with that FDF.
1. A method for managing a distributed Fibre Channel over Ethernet (FCoE) fabric configured to encapsulate Fibre Channel frames in Ethernet frames, the method comprising:
instantiating a first switch link between a controlling FCoE forwarder (cFCF) of the distributed FCoE fabric and a first FCoE data-plane forwarder (FDF), wherein the distributed FCoE fabric further comprises a set of FDFs;
receiving, from the first FDF, a first discovery advertisement message that contains a fabric name field for identifying the distributed FCoE fabric, wherein the fabric name field comprises a first value;
responsive to determining the first FDF is no longer part of the distributed FCoE fabric based on the first value of the fabric name field, de-instantiating the first switch link between the cFCF and the first FDF; and
instantiating a second switch link between the cFCF and the first FDF.
2. The method of claim 1 , wherein the cFCF is configured to de-instantiate the first switch link responsive to not receiving any discovery advertisements from the first FDF within a time period; and
wherein the first discovery advertisement message that contains the first value of the fabric name field is received prior to expiration of the time period.
3. The method of claim 1 , wherein determining the first FDF is no longer part of the distributed FCoE fabric further comprises:
determining the first FDF is no longer part of the distributed FCoE fabric based on reception, from the first FDF, of at least two consecutive discovery advertisement messages having a fabric name set to the first value.
4. The method of claim 1 , further comprising:
prior to receiving the first discovery advertisement message having the first value of the fabric name field, receiving from the first FDF a second discovery advertisement message comprising a second fabric name field that contains a world-wide name (WWN) associated with the distributed FCoE fabric.
5. The method of claim 1 , further comprising:
responsive to determining the first FDF is no longer part of the distributed FCoE fabric based on the first value of the fabric name field, notifying the set of FDFs that the first FDF is no longer part of the distributed FCoE fabric by transmitting a route distribution message.
6. The method of claim 1 , wherein de-instantiating the first switch link between the cFCF and the first FDF further comprises:
transmitting, to the first FDF, a Clear Virtual Links message that specifies the first switch link and associated ports of the first FDF to be de-instantiated.
7. The method of claim 1 , wherein the first value comprises a zero value indicating the first FDF is not part of any distributed FCoE fabric.