OSPF breaks on new Active member after cluster fail-over.
RouteD traces on former Active member (still RouteD master member) show that new Active member (still RouteD slave member) sends OSPF Initial State request. Former Active member (still RouteD master) responds to the request, it is unable to send OSPF protocol state:
Date Time master: INIT_STATE_REQ from Member_ID_of_new_Active, seq=...; local seq=... Date Time CLUSTER: Sending initial state START to Member_ID_of_new_Active Date Time igmp_cluster_send_init_state: initial state d'load to member Member_ID_of_new_Active started Date Time igmp_cluster_send_rstats_state: sending global rstats Date Time cpcl_should_send() returns -1 Date Time igmp_cluster_send_sstats_state: sending global sstats Date Time cpcl_should_send() returns -1 Date Time OSPF CLUSTER: iface X.X.X.X(Name_of_Interface) send <DR> Date Time cpcl_should_send() returns -1 Date Time OSPF CLUSTER: iface X.X.X.X(Name_of_Interface) send crypto sequence <0x...> Date Time cpcl_should_send() returns -1 Date Time OSPF CLUSTER: Master sends neighbor Y.Y.Y.Y(Name_of_Interface) <FULL> Date Time cpcl_should_send() returns -1
Cause
RouteD master member fails to synchronize the OSPF state because this cluster member is not in Running state anymore.