mev-inspect-py

Maximal extractable value inspector for Ethereum, to illuminate the dark forest 🌲💡

Given a block, mev-inspect finds:

• miner payments (gas + coinbase)
• tokens transfers and profit
• swaps and arbitrages
• ...and more

Data is stored in Postgres for analysis.

Install

mev-inspect-py is built to run on kubernetes locally and in production.

Dependencies

• docker
• kind, or a similar tool for running local Kubernetes clusters
• kubectl
• helm
• tilt

Set up

Create a new cluster with:

kind create cluster

Set an environment variable RPC_URL to an RPC for fetching blocks.

mev-inspect-py currently requires a node with support for Erigon traces and receipts (not geth yet 😔).

pokt.network's "Ethereum Mainnet Archival with trace calls" is a good hosted option.

Example:

export RPC_URL="http://111.111.111.111:8546"

Next, start all services with:

tilt up

Press "space" to see a browser of the services starting up.

On first startup, you'll need to apply database migrations with:

./mev exec alembic upgrade head

And load prices data

./mev prices fetch-all

Monolithic (non-kubernetes) install instructions

For an alternative means of running mev-inspect-py for smaller set-ups or debug purposes see the monolithic install instructions.

Usage

Inspect a single block

Inspecting block 12914944:

./mev inspect 12914944

Inspect many blocks

Inspecting blocks 12914944 to 12914954:

./mev inspect-many 12914944 12914954

Inspect all incoming blocks

Start a block listener with:

./mev listener start

By default, it will pick up wherever you left off. If running for the first time, listener starts at the latest block.

Tail logs for the listener with:

./mev listener tail

And stop the listener with:

./mev listener stop

Backfilling

For larger backfills, you can inspect many blocks in parallel

To inspect blocks 12914944 to 12915044, run

./mev backfill 12914944 12915044

This queues the blocks in Redis to be pulled off by the mev-inspect-worker service

To increase or decrease parallelism, update the replicaCount value for the mev-inspect-workers helm chart

Locally, this can be done by editing Tiltfile and changing "replicaCount=1" to your desired parallelism:

k8s_yaml(helm(
    './k8s/mev-inspect-workers',
    name='mev-inspect-workers',
    set=["replicaCount=1"],
))

You can see worker pods spin up then complete by watching the status of all pods

watch kubectl get pods

To see progress and failed batches, connect to Redis with

./mev redis

For total messages, query:

HLEN dramatiq:default.msgs

For messages failed and waiting to retry in the delay queue (DQ), query:

HGETALL dramatiq:default.DQ.msgs

For messages permanently failed in the dead letter queue (XQ), query:

HGETALL dramatiq:default.XQ.msgs

To clear the queue, delete keys for the main queue and delay queue

DEL dramatiq:default.msgs
DEL dramatiq:default.DQ.msgs

For more information on queues, see the spec shared by dramatiq

Backfilling a list of blocks

Create a file containing a block per row, for example blocks.txt containing:

12500000
12500001
12500002

Then queue the blocks with

cat blocks.txt | ./mev block-list

To watch the logs for a given worker pod, take its pod name using the above, then run:

kubectl logs -f pod/mev-inspect-worker-abcdefg

(where mev-inspect-worker-abcdefg is your actual pod name)

Exploring

All inspect output data is stored in Postgres.

To connect to the local Postgres database for querying, launch a client container with:

./mev db

When you see the prompt:

mev_inspect=#

You're ready to query!

Try finding the total number of swaps decoded with UniswapV3Pool:

SELECT COUNT(*) FROM swaps WHERE abi_name='UniswapV3Pool';

or top 10 arbs by gross profit that took profit in WETH:

SELECT *
FROM arbitrages
WHERE profit_token_address = '0xc02aaa39b223fe8d0a0e5c4f27ead9083c756cc2'
ORDER BY profit_amount DESC
LIMIT 10;

Postgres tip: Enter \x to enter "Explanded display" mode which looks nicer for results with many columns.

FAQ

How do I delete / reset my local postgres data?

Stop the system if running:

tilt down

Delete it with:

kubectl delete pvc data-postgresql-postgresql-0

Start back up again:

tilt up

And rerun migrations to create the tables again:

./mev exec alembic upgrade head

I was using the docker-compose setup and want to switch to kube, now what?

Re-add the old docker-compose.yml file to your mev-inspect-py directory.

A copy can be found here

Tear down docker-compose resources:

docker compose down

Then go through the steps in the current README for kube setup.

Error from server (AlreadyExists): pods "postgres-client" already exists

This means the postgres client container didn't shut down correctly.

Delete this one with:

kubectl delete pod/postgres-client

Then start it back up again.

Maintainers

• @lukevs
• @gheise
• @bertmiller

Contributing

Flashbots is a research and development collective working on mitigating the negative externalities of decentralized economies. We contribute with the larger free software community to illuminate the dark forest.

You are welcome here <3.

• If you want to join us, come and say hi in our Discord chat.
• If you have a question, feedback or a bug report for this project, please open a new Issue.
• If you would like to contribute with code, check the CONTRIBUTING file.
• We just ask you to be nice.

Security

If you find a security vulnerability on this project or any other initiative related to Flashbots, please let us know sending an email to [email protected].

---

Made with ☀️ by the ⚡🤖 collective.