Pyth Agent v2.4.4

This release should improve on-chain publishing frequency and can hopefully increase the uptime and reduce the latency of our price feeds.

What's Changed

• fix: fire and forget update tx by @ali-bahjati in #102

Full Changelog: v2.4.3...v2.4.4

Pyth Agent v2.4.3

On this change the defaults have been updated and the logic is slightly improved. exporter.maximum_total_compute_fee_micro_lamports config is removed and exporter.maximum_compute_unit_price_micro_lamports is added. As our parsed is flexible this change won't be breaking.

What's Changed

• refactor: improve price cap configuration by @ali-bahjati in #101

Full Changelog: v2.4.2...v2.4.3

v2.4.2

What's Changed

• chore: add slot gap to pythnet options by @ali-bahjati in #98
• fix: use both latest publisher and aggregate slot for dynamic pricing by @ali-bahjati in #99

Full Changelog: v2.4.1...v2.4.2

Pyth Agent v2.4.1

What's Changed

• fix: use solana-client pubsub lib instead of solana-shadow by @ali-bahjati in #97

Full Changelog: v2.4.0...v2.4.1

Pyth Agent v2.4.0

What's Changed

• Priority fee defaults tuning
• Market hours weekly schedule support - this allows the agent to automatically respect new weekly_schedule metadata if defined on a product. This feature requires no action from configuration/user perspective.

The relevant docker image is available here https://gallery.ecr.aws/pyth-network/agent

Full Changelog: v2.2.0...v2.4.0

Pyth Agent v2.3.1

What's Changed

• Update config.sample.pythnet.toml by @ali-bahjati in #89
• Update config.sample.pythnet.toml by @ali-bahjati in #90
• fix: use publisher latest slot for staleness based pricing by @ali-bahjati in #91

Full Changelog: v2.3.0...v2.3.1

Pyth Agent v2.3.0

What's Changed

• chore: fix typo by @ali-bahjati in #85
• Lower CU budget for mainnet by @guibescos in #86
• Use staleness for pricing by @ali-bahjati in #87
• Update config.sample.pythnet.toml by @hkaliraipyth in #88

Full Changelog: v2.2.0...v2.3.0

Pyth Agent v2.2.0

What's Changed

• fixing readme api link by @ndiakom in #80
• Add minimal docker image to tun pyth-agent by @146karol in #81
• Update config.sample.pythnet.toml by @hkaliraipyth in #82
• Fix documentation link in README.md by @niccoloraspa in #83
• feat: add dyanmic compute unit fee calculation by @ali-bahjati in #84

New Contributors

• @ndiakom made their first contribution in #80
• @146karol made their first contribution in #81
• @niccoloraspa made their first contribution in #83

Full Changelog: v2.1.0...v2.2.0

Pyth Agent v2.0.1

Breaking changes

With this release, the TOML config format abolishes the pythd-compatible key store, in favor of simpler and more readable string pubkey values. This enables us to cleanly distribute sample Pythnet/Pythtest configs with correct program/mapping public keys. Here's how it works:

# Old v1.x.x way
[primary network]
key_store.root_path = "/path/to/keystore"
key_store.publish_keypair_path = "publish_key_pair.json" # Relative path from root_path, "publish_key_pair.json" by default
key_store.program_key_path = "program_key.json" # Relative path from root_path, "program_key.json" by default
key_store.mapping_key_path = "mapping_key.json" # Relative path from root_path, "mapping_key.json" by default

# [...]

# New v2.0.0 way
[primary_network]
key_store.publish_keypair_path = "/path/to/keypair.json" # The root_path is gone, we specify the full path
# Not using separate files anymore
key_store.program_key = "LiteralProgramPubkeyInsideTheConfig" # contents of legacy program_key.json;
key_store.mapping_key = "LiteralMappingPubkeyInsideTheConfig" # contents of legacy mapping_key.json

# [...]

See README for more details and automatic migration instructions. You can also consult config/config.sample.*.toml for reference.

Accumulator publishing

With this release, we begin official support for the Pythnet/Pythtest-crosschain accumulator program. This program is a part of a new approach to cross-chain publishing, which improves throughput and latencies for cross-chain price feed consumers. To enable accumulator support in your agent, add this to your pythnet/pythtest-crosschain config:

[primary_network]

# [...]
# Pythnet/pythtest-crosschain accumulator key
key_store.accumulator_key = "7Vbmv1jt4vyuqBZcpYPpnVhrqVe5e6ZPb6JxDcffRHUM"

# IMPORTANT: Exporter batch size must be decreased to 7 to support
# larger accumulator transactions, when accumulator_key is set. 
exporter.max_batch_size = 7

Other changes

• Sample configs now include the correct program/mapping keys for each network. This includes the accumulator_key option for the new Pythnet/Pythtest publishing approach.
• config/config.toml option ordering has been cleaned up for readibility and correctness.
• Log messages from publisher permission lookup have been clarified

Full Changelog: v1.4.0...v2.0.1

Pyth Agent v1.4.0

What's Changed

• Improved gas efficiency for slow moving prices - Starting with this release, the agent will not publish prices with unchanged price/confidence/status, unless a configurable delay from last publishing has passed or if the price does change. The delay is configured with a new option called <network>.exporter.unchanged_publishing_threshold (at least 5s between unchanged updates by default),
• Moved some non-essential log messages to the debug log level

Docker image

You can find the pyth-agent Docker image here: https://gallery.ecr.aws/pyth-network/agent

Full Changelog: v1.3.1...v1.4.0