Safe Transaction Service

Keeps track of transactions sent via Safe contracts. It uses events and tracing to index the txs.

Transactions are detected in an automatic way, so there is no need of informing the service about the transactions as in previous versions of the Transaction Service.

Transactions can also be sent to the service to allow offchain collecting of signatures or informing the owners about a transaction that is pending to be sent to the blockchain.

Swagger (Mainnet version) Swagger (Göerli version)

Index of contents

Setup for development

Use a virtualenv if possible:

python -m venv venv

Then enter the virtualenv and install the dependencies:

source venv/bin/activate
pip install -r requirements-dev.txt
pre-commit install -f
cp .env.dev .env
./run_tests.sh

Setup for development using docker

docker-compose -f docker-compose.yml -f docker-compose.dev.yml up

Setup for production (event indexing)

Since version 3.0.0 transaction service can be configured to rely on event indexing when SafeL2 version is used. Only contracts from v1.3.0 onwards with L2 events will be indexed.

An example environment file can be used for the L2 setup:

cp .env.l2.sample .env

Edit .env file to add ETHEREUM_NODE_URL (on the example a Polygon public node is used) and remember to modify DJANGO_SECRET_KEY to use a strong key. The rest of the configuration does not need to be modified. Then:

docker-compose build --force-rm
docker-compose up

For more parameters check base.py file.

Setup for a custom network

If the network is not supported yet contracts can be deployed using the deployment instructions and then a PR should be provided adding the deployment block number and the address (address will be the same for every network).
Only ProxyFactory and GnosisSafeL2 must be configured. +L2 must be added to the Safe L2 contract versions, so the service knows the contract can be indexed using events. For us to accept the PR network must be on https://github.com/ethereum-lists/chains .
You can always set this up later using the admin panel if your network is not supported, going to the Master Copies and Proxy Factories.
We recommend using event indexing for every network where transaction fees are not relevant, so a tracing node is not required and everything can be indexed using events with the Safe L2 version.

Setup for production (tracing mode)

This is the recommended configuration for running a production Transaction service. docker-compose is required for running the project.

bash cp .env.tracing.sample .env

Configure the parameters needed on .env. These parameters need to be changed:

DJANGO_SECRET_KEY: Use a strong key.
ETHEREUM_NODE_URL: Http/s address of a ethereum node. It can be the same than ETHEREUM_TRACING_NODE_URL.
ETHEREUM_TRACING_NODE_URL: Http/s address of an OpenEthereum node with tracing enabled.

If you don't want to use trace_filter for the internal tx indexing and just rely on trace_block, set:

ETH_INTERNAL_NO_FILTER=1

For more parameters check base.py file.

Then:

docker-compose build --force-rm
docker-compose up

The service should be running in localhost:8000. You can test everything is set up:

curl 'http://localhost:8000/api/v1/about/'

You can go to http://localhost:5555/ to check the status of the task queue, also you can configure prometheus metrics.

For example, to set up a Göerli node:

Run an OpenEthereum node in your local computer:

openethereum --chain goerli --tracing on --db-path=/media/ethereum/openethereum --unsafe-expose

Edit .env so docker points to the host OpenEthereum node:

ETHEREUM_NODE_URL=http://172.17.0.1:8545
ETHEREUM_TRACING_NODE_URL=http://172.17.0.1:8545

Then:

docker-compose build --force-rm
docker-compose up

Use admin interface

Services come with a basic administration web ui (provided by Django) by default on http://localhost:8000/admin/

A user must be created to get access:

docker exec -it safe-transaction-service-web-1 python manage.py createsuperuser

Safe Contract ABIs and addresses

Service maintenance

Service can run into some issues when running in production:

Indexing issues

You can tell there are indexing issues if:

Executed transactions are missing from the API (all-transactions, multisig-transactions, module-transactions... endpoints). If you use the Safe{Wallet} Web client you should check what is the current state of the Safe Client Gateway cache as it might have outdated data.
Asset transfers (ERC20/721) are missing from all-transactions or transfers endpoints.
You see error logs such as "Cannot remove owner" or similar inconsistent errors when worker-indexer is processing decoded data.

There are multiple options for this. Connect to either web or worker instances. Running commands inside of tmux is recommended (installed by default):

python manage.py check_index_problems: it will try to automatically fix missing transactions. Tokens related transactions (ERC20/721) will not be fixed with this method. This method will take a while, as it needs to compare database data with blockchain data for every Safe.
python manage.py reindex_master_copies --from-block-number X --addresses 0x111 0x222: if you know the first problematic block, it's faster if you trigger a manual reindex. --addresses argument is optional, but if you know the problematic Safes providing them will make reindexing way faster, as only those Safes will be reindexed (instead of the entire collection).

If you see ERC20/ERC721 transfers missing:

python manage.py reindex_erc20 --from-block-number X --addresses 0x111 0x222: same logic as with reindex_master_copies.

FAQ

Why `/v1/safes/{address}` endpoint shows a nonce that indicates that a transaction was executed but the transaction is not shown or marked as executed in the other endpoints?

/v1/safes/{address} endpoint uses eth_call from the RPC to get the current information for a Safe, so there's no delay and as soon as a transaction is executed it will be updated. The other endpoints rely on polling, indexing decoding and processing of traces/events and take longer (shouldn't be more than half a minute).

How do you handle reorgs?

When indexed every block is marked as not confirmed unless it has some depth (configured via ETH_REORG_BLOCKS environment variable). Not confirmed blocks are checked periodically to check if the blockchain blockHash for that number changed before it reaches the desired number of confirmations, if that's the case, all blocks from that block and the transactions related are deleted and indexing is restarted to the last confirmed block.

If I add my chain to safe-eth-py will you support it?

No, for a chain to be supported we need to set up a dedicated infra for that network and have a proper RPC

How can I interact with service?

Aside from using standard HTTP requests:

Safe{Core} API Kit
Safe-eth-py
Safe CLI: It has a tx-service mode to gather offchain signatures.

What chains do you officially support?

https://docs.safe.global/safe-core-api/available-services

What means banned field in SafeContract model?

The banned field in the SafeContract model is used to prevent indexing of certain Safes that have an unsupported MasterCopy or unverified proxies that have issues during indexing. This field does not remove the banned Safe and indexing can be resumed once the issue has been resolved.

Troubleshooting

Issues installing grpc on a Mac M1

If you face issues installing the grpc dependency locally (required by this project) on a M1 chip, set GRPC_PYTHON_BUILD_SYSTEM_OPENSSL=1 and GRPC_PYTHON_BUILD_SYSTEM_ZLIB=1 and then try to install the dependency again.

Contributors

See contributors

Zebec-protocol/safe-transaction-service