TIPC's algorithm for identifying on-node and off-node destinations that overlap a multicast name sequence range is not as efficient as it could be. Rather than traversing the list of all known name publications within the cluster, it should just traverse the (potentially
much shorter) list of name publications made by the node itself, and determine if any off-node destinations exist by comparing the sizes of the two lists. (Since the node list must be a subset of the cluster list, a difference in sizes means that at least one off-node destination must exist.)
Patch for TIPC 1.7.5
Logged In: YES
user_id=995681
Originator: YES
A patch for TIPC 1.7.5 is now available that fixes this issue.
File Added: 0010-Optimization-to-multicast-name-lookup-algorithm.txt