Reimplement the forget(delete) in store and connect these functions through dht.

So apparently Linux doesn't like when we open too many UDP connections. Currently we're opening one connection per dial (for each send, that is).

Instead, we should re-use the same connection for both the listener and the sender.

It seems that the republishHandler in store only iterates over the local items (db.localItems), but according to:

The whole database must be replicated every hour?

Ping @Hjortsberg

Marshal/unmarshal a Protobuf message between two nodes.

• Receive message from CLI
• Make relevant call to routing table
• Implement first stage before sending of data to network and store packages

We need to be able to send strings to the Kademlia network via REST.

I'm not 100% sure how this one would be solved.

We'd get a dependency cycle if the routing table calls any function in the DHT. So this logic would probably be added to the DHT and not (r *Table).Add.

One way to do it:

1. DHT gets a message from node A.
2. DHT checks if the bucket is full, with a new route.Table method, i.e.
   (r *Table).Full(id node.ID) bool.
3. DHT gets a...
   a. true: Go to 4.
   b. false: Add new contact A, and go to 6.
4. DHT retrieves the contact at the head of the bucket (least recently seen, B), using a new method (r *Table).Head(id node.ID) route.Contact
5. DHT pings the least recently seen node, B, and if it gets a...
   a. Response: Ignore new contact A.
   b. No response: Add new contact A (evicting node B).
6. Done.

Logging slow, ok?
Make faster, ok?
Zero allocations, nice!
Fix, ok?

This one could be removed now: https://github.com/optmzr/camomile/blob/master/cmd/udptestclient/udptestclient.go

Probably just a simple hash table?

Spawn a go routine, and a channel.

Implement these messages:

message Message { MessageID : Bytes }
message Ping { Message* }
message Pong { Message* }
message Store { Message* Value* }
message FindNode { Message* NodeID : 160 bit }
message NodeInfo { Message* IP : bytes Port : uInt NodeID : bytes }
message FindValue { Message* Key : bytes }
message Value { Message* Key : bytes Value : UTF-8 }

• https://google.github.io/flatbuffers/
• https://developers.google.com/protocol-buffers/
• https://msgpack.org/

Identify the bug where many (10+) nodes sometimes reports this error:

2019/09/29 10:27:47 Channel not found in table

The required content of the lab report is described in the Workflow section in the lab assignment text. For example:

• draft architecture description
• initial backlog, prioritized with objective to first reach a minimal working solution
• solution outlines for your prioritized items on the backlog and objectives in the lab assignment

The report shall be uploaded at midnight 2 days before your review slot (so reviewers will have one day to check it). Book your time slot with a supervisor for review. The link for booking a time slot is here:

Implement ping so that you can ping from the CLI to a node by sending Protobuf packets.

Example of format usage: ./dhtctl -address <ipv6ip> -put "UTF8 string"

-address: string
    - put: string (value) ret:string (key)
    - get: string (key)    ret:string (value)
    - exit: bool

We've missed this one:

Add the necessary components as decided in the serialization issue.