myneworder/fault-tolerance-demo

DEPRECATED: A demo of tendermint's fault tolerance

fault-tolerance-demo

A demo of tendermint's fault tolerance.

Start a testnet with four nodes, monitor it to ensure it's healthy (making blocks). Kill a node, the network should remain healthy. Kill a second node, the network should cease making blocks. Bring a stopped node back online and the network should continue making blocks.

Pre-requisites

• Install tendermint, mintnet (ports branch), tmsp dummy, tmsp-cli, netmon
• Launch machines test1, test2, test3, test4
• jq for pretty printing json

Scripts

• launch.sh - launch the testnet on the machines and start the netmon to monitor them
• rm.sh - remove the testnet and all created files. inverse of launch.sh
• start_mint.sh - start a local tendermint node that syncs with the testnet. acts as a proxy for transactions
• send_tx.sh - send thousands of random transactions to the local node to broadcast to the network

Commands

• curl -s http://localhost:46670/get_chain?chainID=\"mynewchain\" | jq . - monitor the health and current state of the chain
• curl -s http://localhost:46657/unconfirmed_txs | jq . - see the txs in the mempool
• curl -s http://localhost:46657/unconfirmed_txs | jq .result[1].n_txs - size of the mempool
• mintnet rm --force --machines "test1" mynewapp - kill a node
• mintnet start --machines "test1" --seed-machines "test[1-4]" mynewapp mynewchain - start a node

Notes

Send transactions with

curl http://localhost:46657/broadcast_tx_sync?tx=\"$(toHex foo=bar)\"

where toHex is the bash function

function toHex() {
	echo -n $1 | hexdump -ve '1/1 "%.2X"'
}

And query for the result with

tmsp-cli query foo

which should, in this case, return bar