Skip to content

Latest commit

 

History

History
132 lines (107 loc) · 7.32 KB

CHANGELOG.md

File metadata and controls

132 lines (107 loc) · 7.32 KB
Raw

Changelog

All notable changes for substrate-archive will be documented in this file.

The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.

[Unreleased]

Added

  • RabbitMQ is used by-default for task-queue storage. This is an extra dependency that must be started as a service outside of Archive.

Changed

  • BREAKING: use RabbitMq instead of Postgres for the background tasks queue. Migrations will take place automatically.
    • table _background_tasks will be dropped.
    • table _sa_config will be added.
  • Storage is inserted concurrently based on idle SQL connections.

Removed

  • BREAKING Dispatch generic on Archive and ArchiveBuilder.
  • remove native execution. All runtimes will be executed in WASM, either interpreted or compiled.

[v0.6.0] - 2021-06-24

Added

  • Use async-std instead of smol for async tasks. (#301)
  • Additional tests for missing storage & pagination query. (#254)
  • Storage Chain Support (#292)
  • v0.9.5, v0.9.6, v0.9.7 wasm tracing runtimes

Changed

  • BREAKING: PostgreSQL _background_tasks table stores data as jsonb rather than bytea. Upgrading should happen automatically when running v0.6.0, but you will not be able to revert to any previous version of archive. (#254)
  • When restoring storage, blocks are paginated using the max_block_load parameter. This cuts memory usage down when many blocks are missing from storage. (#254)
  • Update dependencies to match client 0.9.7.

Removed

  • BREAKING db_actor_pool_size is no longer configurable on the builder. Actor pool no longer exists. (#254)
  • BREAKING max_tasks from builder. Max tasks will always be the number of CPUS on the system. (#301)
  • BREAKING task_workers from builder. Task workers will always be number of CPUS on the system. (#301)

Fixed

  • archive no longer hangs on Ctrl-C. (#301)

[v0.5.2] - 2021-06-02

Added

  • Test for tracing enabled wasm-blobs v0.9.0, v0.9.1, v0.9.2, v0.9.3 (#284) (cd6a446)
  • Unit testing CI workflow (#288) (482af68)

Changed

  • Clarify release checklist (#279) (9abef6e) && (#288) (482af68)
  • Update dependencies to match runtime 0.9.3.
  • Refactor tracing to work with the latest tracing changes in substrate (#273) (b322ded)

Fixed

  • Re-Compile WASM Runtime v0.8.30 with rust compiler version nightly-02-27-2021 to fix 'Storage Root Mismatch' when syncing.
  • Fixed wasm runtimes so their names are uniform.

[v0.5.1] - 2021-05-06

Added

  • Release checklist

Changed

  • Pinned to substrate/polkadot release v0.8.30
  • Keeping the substrate-archive and substrate-archive-backend crates's versions aligned.

[v0.5.0] - 2021-03-29

Added

  • ArchiveBuilder struct for constructing the indexer.
    • returns a trait, Archive that manages the underlying actor runtime
  • Archive now reads the CHAIN_DATA_DB environment variable if the path to the backend chain database is not passed directly.
  • max_block_load to configure the maximum number of blocks loaded at once
  • Two new options to the configuration relative to state tracing:
    • targets for specifying runtime targets to trace in WASM
    • folder where WASM blobs with tracing enabled are kept.
    • More on state-tracing here
  • Runtime versions between blocks are cached, speeding up block indexing.
  • Notification stream from Postgres indexes storage changes in the background.

Changed

  • no longer need to instantiate a client and manually pass it to the Archive
  • rename run_with to drive
  • Archive now accepts a postgres URL instead of a postgres URL split into its parts.
  • Archive will take Postgres URL from environment variable DATABASE_URL as a fallback.
  • Use SQLX for migrations
    • SQLX will create a new table in the database, sqlx_migrations or similiar.
    • the older table, __refinery_migrations can be safely dropped
  • Print out the Postgres URL at startup
  • config is now separated into sections for readability. Migration is manual but looking at the new archive.conf in polkadot-archive or node-template-archive folders should help.
  • Postgres SQL queries are now type checked
  • Refactor file layout to substrate-archive and substrate-archive-backend.
  • Decouple Database actors
  • upgrade to SQLx 0.5.0
  • Overhaul of block indexing. Now uses a Iterator to only collect batches of blocks from the database, taking advantage of the better reading performance of sequential data access. Gathering blocks by RPC is no longer done.
  • speed up query for difference between the storage and blocks table -
  • switch to a leaner actor framework, xtra
  • switch to a persistant background-task-queue for executing blocks, coil.
  • batch inserts no longer starve postgres pool of connections
  • Switch substrate-archive LocalCallExecutor with Substrate's LocalCallExecutor

Removed

  • the last frame dependency, frame-system. Archive now relies only on generic traits defined in substrate-core.
  • RPC URL. An RPC URL is no longer required to function.

[v0.4.0] - 2021-01-24

Added

  • integrate database migrations into substrate-archive library

Changed

  • Speed up Storage Indexing by re-executing blocks
  • refactored error types

Removed

  • dependencies for service and backend

Fixed

  • storage inserts

[v0.3.1]

Added

  • node-template-archive

[v0.3.0]

Added

  • Create a CLI for indexing kusama
  • New PostgreSQL Schema
  • Actors to model dataflow

Changed

  • Use a rocksdb-backed substrate client instead of RPC for indexing