Release Notes - v0.48.0

このコンテンツはまだ日本語に翻訳されていません。

v0.48.0 (May 26, 2026)

Title: More resilient Mithril downloads, tunable backfill behavior, and safer live sync near the chain tip

Date: May 26, 2026

Version: v0.48.0

Hi folks! Here’s what we shipped in v0.48.0.

Added inspect Mithril sync with a dedicated debug endpoint: Operators can expose a separate debug endpoint during Mithril sync to investigate node activity more easily.

Improved tune history import pacing to fit each deployment more precisely: Operators can set the backfill batch size through configuration or flags to better match local storage and workload needs.
Enhanced recover stalled Mithril downloads with less wasted progress: Dingo now detects paused snapshot downloads, validates retry settings early, and resumes partial transfers instead of starting over.
Refined reduce memory pressure during dense history imports: Dingo now writes accumulated history data sooner during heavy imports, which keeps large catch up runs steadier.

Corrected keep live sync moving when the node reaches the latest block: Nodes now avoid a timing issue at the latest known block that could leave progress waiting for a later update.

Clarified configure backfill batch size when history import needs different pacing: Operators can use the new backfill batch size setting in configuration or flags when a smaller or larger block window fits the environment better.
Highlighted use Mithril idle retry controls when downloads pause without finishing: Mithril sync now supports settings that detect paused downloads, control retry behavior, and continue from partial archive data.
Emphasized open a separate debug endpoint during Mithril troubleshooting: Mithril sync can expose a debug endpoint on a separate port when deeper inspection is needed.
Summarized expect steadier history import behavior during dense catch up work: Dingo now writes accumulated history data earlier during heavy batches to reduce memory pressure.
Noted expect smoother live follow behavior at the latest known block: Nodes now continue waiting for the next block more reliably when sync reaches the current tip.

Thank you for trying!