millerm / nabu

A minimal Java implementation of IPFS

Investigating tracing for IPFS with a minimal implementation

This project explores adding tracing functionality to IPFS. To create a prototype, we started with Nabu, a minimal implementation of IPFS.

To focus on the tracing aspect, several features of Nabu were disabled:

• Bootstrap discovery: Nabu doesn't include bootstrap peer discovery. To establish a private network, ten peers are manually connected to each other.
• Optimistic BitSwap and flow control: These features, related to data exchange between peers, were removed for the prototype.
• File caching: The ability to cache fetched files was also disabled.

Note: Only the content retrieval (GET) path is traced.

Setup

./compile.sh

./run-local.sh <node-id>

node-id can be 0-9.

API

PUT

Description

Adds content provided in raw request body and returns the content id for it.

PUT /api/v0/block/put

Response

{
  "cid" : string
}

Status Codes

• 200 - OK
• 400 - BAD_REQUEST

GET

Description

Retrieves content provided for the provided cid. Optionally exports trace for the request if trace is set to 1.

GET /api/v0/block/get?cid=<cid>[&trace=1]

Response

Response body contains content bytes.

Status Codes

• 200 - OK
• 400 - BAD_REQUEST

Example

Request 1

curl -X PUT -d "abc" http://localhost:5001/api/v0/block/put

Response 1

{"Hash":"bafkreif2pall7dybz7vecqka3zo24irdwabwdi4wc55jznaq75q7eaavvu"}

Request 2

curl 'http://localhost:5001/api/v0/block/get?cid=bafkreif2pall7dybz7vecqka3zo24irdwabwdi4wc55jznaq75q7eaavvu&trace=1'

Response 2

abc