Multi Node Tracing

[andytill]

Allow erlyberly to connect to multiple nodes, when a trace is applied, apply it to all nodes.

• Show a list of nodes in the connection window
• Toggle traces to all connected nodes
• Show errors per node, for example when a trace cannot be applied because the module doesn't exist.
• The module list should show modules from all nodes.
• When a module exists in only one node, and than node disconnects, it should be removed, or maybe just disabled.
• Have a drop down in the disconnect button, to show multiple nodes and disconnect from them individually
• Allow connections to nodes while already connected to other nodes.
• Processes should be shown for one node, selected from a drop down.
• VM memory and xref should be disabled when more than one node is connected.

[ruanpienaar]

This one seems interesting

[andytill]

Alternatively we just use the multi node tracing in dbg:m/1 and have the concept of a "primary" traced node which will be queried for functions and other data, and "secondary" nodes where the traces will also be applied. The only big problem I see is that all nodes would need to be clustered for it to work, though erlyberly could check if they are clustered by querying erlang:nodes([connected]) on each node it is tracing on.

Need to investigate more but this seems like 20% of the work for 80% of the benefit.