Collect Testnet On-Chain Data

[gheise]

To understand at validator behavior, block hashes and additional on-chain data needs to be collected to assess the ~perfomance of the relay~ rate of slashing and calculate validator rewards

• [x] Create tables for on-chain testnet data on the testnet database ~- [x] Write on-chain testnet data to new tables by running script on EC2 with write permissions~

We've heavily relied on scripts reading from db + access controls to integrate on-chain and relay data, but this has several flaws:

1. Not scalable: as the number of data sources grows, writing scripts and running them on EC2 will create a large maintenance and documentation requirement, requiring lots of resources from data and devops to monitor.
2. Error prone: increases the bug surface area of integrations directly connected to the relay db and internal metrics service like metabase

A better solution would be to use AWS Glue, an integrated AWS service for ETL, where scripts that read on-chain data can transform data from the relay db upon a trigger, like a new payload delivery, gathering on-chain data and writing to separate tables on the relay database without the overhead of maintaining scripts scattered in EC2 instances or altering the relay setup

Update

With the help of many we've decided to go with Apache Airflow, managed by Amazon. The repository for the graphs that will replace our internal scripting with regards to data collection can be found here

[metachris]

I don't fully understand the goal here.

Please define "to assess the perfomance of the relay."

[gheise]

I don't fully understand the goal here.

Please define "to assess the perfomance of the relay."

I clarified above based on our conversation, thanks 😄

[metachris]

how do you calculate validator rewards? do you take into account that builders set their own feeRecipient and send the value in a transaction to the proposer feeRecipient that's sent outside of the chain to the builders?