Skip to content

Latest commit

 

History

History
90 lines (53 loc) · 4.83 KB

README.md

File metadata and controls

90 lines (53 loc) · 4.83 KB
Raw

Summary

This repository contains the PoktInfo services that depend on Pocket's chain data. Rewards info ingests the proofs and claims in order to index every time a node earns by documenting reward and relay amounts. Nodes Info indexes the network's nodes information over time. Location info indexes where the nodes in nodes info are geographically located and other related information such as ISP and IP.

Installation

  1. Follow the Installation steps of common, a requisite for PoktInfo which contains the ORM, generic interactions, and more.
  2. Clone this repository
  3. Follow the steps for the desired service below

Rewards Info

To run this service and gather historical data:

python run_rewards.py history <START_HEIGHT> <END_HEIGHT>

To run this service "live" and collect current and future data:

python run_rewards.py live

run_rewards.py is the main file which invokes the logic in rewards_calc.py.

Logic

Historical:

  1. run_rewards.py invokes run_rewards() which uses process pools to call record_rewards() for each block within the specified range
    1. Calls get_relays_wrapper()which queries all the claims on the state at height - 1 and passes them to get_relays() which matches the proofs from height to the claims and multiplies them by the correct relaysToTokensMultiplier and validatorPercentage
    2. Perform sanity check to ensure that (totalSupply(height) - totalSupply(height - 1)) * validatorPercentage == totalRewards(height)
    3. Saves reward information in db

Live:

Performs the same process as historical mode, but calls record_rewards starting at the last cached block sequentially. If the last cached block is equal to the current height, the process will sleep until the current height increments. We separate historical and live processes because historical allows for indexing multiple blocks at once which is beneficial given Pocket's long query times.

Schema

Please see here for the rewards info schema definition.

Nodes Info

To run this service in historical mode:

python run_rewards.py history <START_HEIGHT> <END_HEIGHT>

To run this service in live mode:

python run_rewards.py live

run_nodes.py is the main file which invokes the logic in nodes_info.py.

Logic

Historical:

  1. run_nodes_info(from_height: int, to_height: int) is called and sequentially calls record_nodes_info_wrapper for each block
    1. The calls are made in series because the state of height - 1 is required to manage changes to a node at height
  2. Queries all nodes at height and iterates through each node to check for changes to the nodes service_url, chains or unstaking_time.
  3. If there is a change, the end_height of the node in the table will be marked to height - 1, and it will create a new row with start_height = height

Live:

Live mode starts from the last indexed height and will call record_nodes_info_wrapper until the indexed height is equal to the current height. It will sleep until the current height increments.

Schema

Please see here for the nodes info schema definition.

Location Info

To run location info:

python3 location_service.py <RAN_FROM>

ran_from specifies the region where the instance of location info is running from. Because of modifications like geomesh, where operators are able to have multiple clients per address by using a regional load balancing service, one must run location info in several regions in order to detect all the actual instances of a given servicer/validator. Based on where this service is ran, you will get different results. (e.g. If you run from the U.S, you will find U.S versions of the network's nodes rather than an exhaustive list) Options: na, eu, sg

Logic and main is in location_service.py.

Be sure that you update the ip-api key in common repo ip_api_utils.py.

Logic

Live:

  1. Every 6 hours run_location_service() is called
    1. Active nodes are queried from nodes_info
    2. For each node, location data is queried from ip-api using the node's service_url
    3. Ifaddress, city, ip or isp differ for the recorded node, than the old one's end_height will be specified and a new row will be created. The ran_from column will be set to the value specified in the CLI argument.

Because you cannot find out where something was physically located in the past, location info only has a live mode.

Schema

Please see here for the nodes info schema definition.