Babylon Benchmark

This repository contains tools and scripts for benchmarking, profiling and generating data in the Babylon blockchain node.

Getting Started

The repository consists of two independent parts, data generation CLI (dgd) and scripts used to demonstrate sync from a snapshot.

The data generation CLI (dgd)

Data generation CLI is built partly from the Babylon auxiliary programs (vigilante, btc-staker, finality provider) and is able to create activated delegations and send finality votes, basically simulating what auxiliary programs are doing.

It starts bitcoind, node0 and node1 docker containers, where node0 and node1 are Babylon validator nodes, node0 is used for all other RPC related queries while node1 is strictly used for submitting delegations.

Building the CLI

1. Build the CLI by running make build.
2. Confirm that dgd CLI has been built successfully by running ./build/dgd version

Command examples

The following command starts:

• 20 finality providers
• 200 btc stakers
• total-delegations - indicates that the daemon will shut down after this number

./build/dgd generate --total-fps 20 --total-stakers 200 --total-delegations 1250000 --babylon-path /data/babylon-benchmark

Sample log from the dgd:

🔎 Found 1 non-finalized block(s). Next block to finalize: 2723
📌 Sending checkpoint for epoch 141 with proof 2
📄 Delegations sent: 22585, rate: 8.00 delegations/sec, ts: Fri Dec  6 15:58:16 UTC 2024, mem: 5347 MB
⏱️ Average delegation submission time: 2.0138 seconds
✍️ Fp e5342f02b71cabec2ef96d8adf1d4d1c98706e3f928911b00b74fed763b98363, voted for block range [2723-2723]
✍️ Fp 63bc13483ff1b3b436ec8a1a4d8c542e4b63fee4f3e3e19e0e7cf3b583b06fee, voted for block range [2723-2723]

Starting main & follower nodes from a snapshot

Follow these steps to set up and run the benchmark:

1. Clone the repository:

   git clone [email protected]:babylonlabs-io/babylon-benchmark.git
   cd babylon-benchmark

2. Initialize, update, and check out the correct version of submodules:

   git submodule init
   git submodule update
   cd submodules/babylon
   git checkout v0.18.0
   cd ../..

Note: By default, the submodules are pointing to version v0.18.0, which corresponds to the Phase 2 devnet-1 snapshot. Always ensure that the Babylon version matches the snapshot you're using. You can verify the correct version in the devnet-k8s repository. If you're using a different snapshot or version, make sure to checkout the appropriate tag or commit in the Babylon submodule.

3. Prepare the snapshot:

   • Download the Phase 2 devnet snapshot from this thread.
   • Place the downloaded .tar.gz file in the snapshots/ folder of your project.
   • The extraction and booting process will be automatically handled by the program. You only need to ensure the .tar.gz file is present in the snapshots/ directory.
   • Update the SNAPSHOT_FILE environment variable in your docker-compose.yml to match your snapshot filename:

     environment:
       - SNAPSHOT_FILE=your_snapshot_name.tar.gz

4. Start the benchmark:

   make start-benchmark-from-snapshot
   

   This command will:

   • Build the Docker images
   • Start both the master and follower nodes
   • Bootstrap the master node with the provided snapshot
   • Start the follower node which will begin syncing with the master
   • Start the profiler once the follower node has started syncing

   The profiler will by default run for 60 seconds. To modify the profiling duration:

   • Open scripts/run-profiler.sh
   • Locate the start_profiling function
   • Change the seconds=60 parameter in the curl command to your desired duration

   For example, to profile for 120 seconds, you would modify the line to:

   curl -o "$PROFILE_FILE" http://localhost:6061/debug/pprof/profile?seconds=120
   

   After completion, the profile data will be automatically saved in the outputs/ folder. You can visualize the profile data using the following command:

   go tool pprof -http=:8080 outputs/profile_<timestamp>.pprof
   

5. Stop the nodes: When you're done with the benchmark, you can stop the nodes using:

   make stop-benchmark
   

   This will shut down both the master and follower nodes and clean up the data directories.

Directory Structure

• snapshots/: Place your blockchain snapshots here.
• scripts/: Contains initialization and profiling scripts.
• outputs/: Contains the profile data.

Notes

• Ensure you have sufficient disk space for the blockchain data and snapshots.
• The profiler will automatically stop once the follower node has synced with the master.
• Review the scripts/init_master.sh and scripts/init_follower.sh for any additional configuration needed.

Troubleshooting

If you encounter any issues:

• Check that the snapshot path in init_master.sh is correct.
• Ensure Docker is installed and running on your system.
• Verify that the required ports are not in use by other applications.

For more detailed information, refer to the individual script files.