There are cases where messages arriving to the OMW Management Server contain both the IP address and the node name of the generating node.
Let's consider a situation as follows: a single MoM Cluster which communicates with four OMW Management Servers. Each OMW Management Server is in a separate network.
Child OMW Server A might be monitoring a node with IP Address 184.108.40.206 and Child OMW Server B might also be monitoring a node with IP Address 220.127.116.11; each OMW is monitoring and receiving the alerts normally on its own, but the problem happens when it tries to forward this data up to the MoM OMW Server: it finds that there are two nodes with the same IP Address of 18.104.22.168 so it drops the message. This happens because OMW does not allow two nodes with the same IP address in its model.
The best way to resolve the proposed the scenario where the same IP address come from different nodes in different networks to a single OMW MoM, would be to configure the nodes with their FQDN and make sure that the agent ID is set so that the messages match via agent ID, but this solution wouldn't work when the the issue happens with SNMP nodes where no agent ID is involved, so the IP address is the preferred way to map the messages in such cases.
The OMW side mapping of message onto a node always depends on which message properties are set in the incoming message.
For non-proxied messages, if both node name and IP address are set, then the validation checks whether the node name maps to the IP address and IP address maps to node name. If yes, it is validated whether the node is configured in OMW’s node bank. It is fully intended that IP addresses and node names are cross-checked if both are specified. If an IP address does not fit to the specified node name or vice versa, then there is no way to find out which item is the correct one, so the message cannot be mapped uniquely to a node. An option would be to just go directly and map using the node name by introducing a new server configuration option.
For proxied messages, if the agent ID of the proxied node is not set, then the validation is done by using the IP address of the managed node. If the node name of an agentless node is contained in the message (but there is no IP address) then the node name is used. If the IP address is contained but there is no node name, then the IP address is used. If both node name and IP address are contained, then a cross-check is done in the same way as for non-proxied messages. But the mapping could be based on node name first if available, and the last resort would be using the IP Address to avoid confusion as the scenario described above.
Therefore a new server configuration option should be introduced which allows new incoming messages to be mapped to the node name if both node name and IP address are contained in the messages.
If message can be validated by using the node name only then IP address should not be checked and no reverse resolution should be performed.
This new configuration option would prevent messages to be discarded in environments where the same IP address is assigned to multiple nodes.