chore(release): prepare v1.0.1

[ebfull]

Motivation

First patch release after v1.0.0. Since the tag, main has accumulated operator-visible fixes (header-sync stall and correctness fixes, VCT validation hardening, block-sync memory bounding, mempool and Sapling verification performance) that should ship as v1.0.1.

Solution

Release prep per .github/PULL_REQUEST_TEMPLATE/release-checklist.md, with crate versions advancing independently (changed crates only):

Crate	1.0.0 →	Why
zakura-network	2.0.0	cargo semver-checks: 4 breaking findings — removed/renamed public items (#222, #154)
zakura-state	2.0.0	DerefMut impl removed from CheckpointVerifiedBlock (#208 documents it as source-breaking; below semver-checks' lint coverage, which reports clean)
zakura-consensus	2.0.0	state service types are in its public init bounds — public-dependency major
zakura-rpc	2.0.0	state/network/consensus service types in its public server signatures — public-dependency major
zakura-chain	1.1.0	additive only: Block::attributed_memory_size_bytes (#159)
zakura-utils	1.0.1	its zakura-rpc requirement moves to 2.0.0 (private dep of the tools)
zakura	1.0.1	the node package; drives the v1.0.1 tag

The remaining six crates are unchanged since v1.0.0 and are not bumped or published.

• Root CHANGELOG.md: new [1.0.1] section. This also relocates the three ### Fixed entries that fix(state): stop serving fabricated commitment metadata #133/fix(sync): retry missing header ranges indefinitely #138 appended under [1.0.0] — both PRs merged after the tag (their changes ship in v1.0.1, and the [Unreleased] section didn't exist yet when they landed), so [1.0.0] returns to the plain "Initial release" entry per CHANGELOG_GUIDELINES.md.
• Crate changelogs for the seven bumped crates, from cargo semver-checks / cargo public-api diff latest.
• Installer pinned to the v1.0.1 tag with the placeholder-zero checksum (release CI substitutes the real value into the published asset copy); Docker pins move to zakuracore/zakura:1.0.1 / zcashd-compat-1.0.1. README --tag line via cargo release replace.
• Retires the upstream-era stored config corpus in its own commit (51 snapshots inherited from Zebra, v1.0.0-beta.* through v5.0.0*, plus feature variants): no Zakura user ever ran them — Zakura's first release is v1.0.0 — the acceptance tests only iterate the directory and look up v{CARGO_PKG_VERSION}.toml by name, and upstream's ancient v1.0.1.toml squatted on this release's snapshot filename. Keeps v1.0.0.toml and custom-conf.toml (loaded by path in test-docker.yml). Zakura's rc-era snapshots go with the rest, per the rc-artifact sweep — shout if you'd rather keep those. (Supersedes test(zakurad): remove stale upstream stored configs #233, closed in favor of a single PR.)
• Regenerated stored config snapshot zakurad/tests/common/configs/v1.0.1.toml (the last_config_is_stored acceptance test derives the filename from the package version).
• ESTIMATED_RELEASE_HEIGHT 3,413,500 → 3,415,700 (tip 3,415,571 on 2026-07-17 per Blockchair, same-day cut), and EOS_PANIC_AFTER 7 → 18 days (CHANGELOG_PARAMS.md row included): the 7-day window would halt every 1.0.1 node ~4 days before the expected Ironwood activation (3,428,143, ~2026-07-28). The halt now lands one week after activation (~2026-08-04, height ~3,436,436), so 1.0.1 spans the upgrade and still forces migration to the post-Ironwood release. Warnings keep their 3-day lead (start ~2026-08-01). Re-estimate if dispatch slips.
• Checkpoints: not refreshed (explicit decision, matching v1.0.0). The lists top out at 3,358,006; the automated checkpoint-update pipeline is gated on the GCP integration-tests workflow, which hasn't produced artifacts since 2026-06-23.

Merging this PR does not cut the release: the create-release dispatch and the release-environment approval remain the controls, and no crates.io publish happens as part of this change.

Testing

• make pre-release RELEASE_TAG=v1.0.1 BASE_TAG=v1.0.0 (output in a PR comment)
• cargo semver-checks per changed library crate: network fails 4 major checks (matching the table); state/chain/consensus/rpc report "no semver update required" — state's break and chain's addition sit outside the tool's lint coverage, hence the manual classification above
• cargo test --release -p zakura --test acceptance -- config_tests (covers last_config_is_stored and stored_configs_work)
• cargo metadata --locked resolves, so the lockfile matches the new graph
• bash -n scripts/install-zakura.sh

Specifications & References

• .github/PULL_REQUEST_TEMPLATE/release-checklist.md
• CHANGELOG_GUIDELINES.md
• chore(release): prepare v1.0.0 #194 (the v1.0.0 prep this mirrors), test(zakurad): remove stale upstream stored configs #233 (superseded stored-config cleanup)

Follow-up Work

• Dispatch create-release for v1.0.1 after merge (separate decision, per the release checklist).
• Publish the seven bumped crates in checklist order (zakura-chain, zakura-state, zakura-consensus, zakura-network, zakura-rpc, zakura-utils, zakura), then promotion, signing, and the installer-metadata PR per the checklist.
• The v1.0.0 GitHub release description inherited the fix(state): stop serving fabricated commitment metadata #133/fix(sync): retry missing header ranges indefinitely #138 entries relocated here — edit it to match the corrected [1.0.0] section.

AI Disclosure

Prepared with Claude Code (version bumps via cargo release, changelog and metadata edits, preflight runs); human review required before merge and dispatch.

🤖 Generated with Claude Code

[ebfull]

Changelog review trail, per the release checklist ("put the list of deleted changelog entries in a PR comment"). Since Release Drafter had no draft for this range, the [1.0.1] section was composed from git log v1.0.0..HEAD (25 commits); these are the changes excluded as internal-only per CHANGELOG_GUIDELINES.md:

Change	Why excluded from the root changelog
#224, #222, #154	header-sync/block-sync internal refactors; their public-API consequences are recorded in zakura-network/CHANGELOG.md [2.0.0] instead
#204	gates raw RocksDB test helpers to cfg(test); no public API or operator-visible behavior
#206	build-script-only removal; recorded in zakura-rpc/CHANGELOG.md [2.0.0]
#210	test-only (serializes legacy reorg replacement e2e tests)
#213	docs lint allowance for generated uint code
#205	devops dependency group bumps
#214	deploy tooling alert format (not a zakurad change)
#197	release automation (v1.0.0 installer metadata)

Everything else since the tag has an entry under [1.0.1] (root) and/or the relevant crate changelog. The four un-numbered commits from the rebase-merge of #202 are linked as #202.

[ebfull]

Output of make pre-release RELEASE_TAG=v1.0.1 BASE_TAG=v1.0.0 on the release commit 13b7d2de8:

==> [1/2] Checking release version

make pre-release-version RELEASE_TAG="v1.0.1" BASE_TAG="v1.0.0" PRE_RELEASE_WARN_CRATE_VERSION_BUMPS="1"
make[1]: Entering directory '/home/sean/build/zakura/zakura'
./scripts/check-release-version.sh "v1.0.1"
Release tag v1.0.1 matches the 'zakura' package version (1.0.1).
Changed publishable crate: zakura (1.0.0 -> 1.0.1)
Changed publishable crate: zakura-chain (1.0.0 -> 1.1.0)
Changed publishable crate: zakura-consensus (1.0.0 -> 2.0.0)
Changed publishable crate: zakura-state (1.0.0 -> 2.0.0)
Changed publishable crate: zakura-network (1.0.0 -> 2.0.0)
Changed publishable crate: zakura-rpc (1.0.0 -> 2.0.0)
Changed publishable crate: zakura-utils (1.0.0 -> 1.0.1)
Crate version bump advisory found no unchanged versions across 7 changed published crate(s).

NOTICE: The working tree has local changes.
        This advisory compared those changes against v1.0.0.
make[1]: Leaving directory '/home/sean/build/zakura/zakura'

Package archive sizes:
  zakura-test-1.0.0.crate: 1718344 bytes
  zakura-tower-fallback-1.0.0.crate: 12042 bytes
  zakura-jsonl-trace-1.0.0.crate: 9582 bytes
  zakura-chain-1.1.0.crate: 1494545 bytes
  zakura-tower-batch-control-1.0.0.crate: 27948 bytes
  zakura-node-services-1.0.0.crate: 36796 bytes
  zakura-script-1.0.0.crate: 46280 bytes
  zakura-state-2.0.0.crate: 539929 bytes
  zakura-consensus-2.0.0.crate: 161708 bytes
  zakura-network-2.0.0.crate: 953642 bytes
  zakura-rpc-2.0.0.crate: 266218 bytes
  zakura-utils-1.0.1.crate: 65175 bytes
  zakura-1.0.1.crate: 488745 bytes

All 13 crates package cleanly.
make[1]: Leaving directory '/home/sean/build/zakura/zakura'

(The version-gate and advisory section above is from the pre-commit run — the packaging step re-ran on the clean release commit; both passed. The NOTICE: working tree has local changes line reflects the pre-commit run only.)

[ValarDragon]

Hrmm is this really 2.0?

[ValarDragon]

I don't think those pub api's should be for public consumption. If we follow exact strictness like this, we'll probably be at 10.0.0 in a few weeks

[ebfull]

In general there might be situations where users will build and cargo will select older (semver compatible) versions in lockfiles or caches, but those would be incompatible.

[ebfull]

We can address this before the version numbers get out of hand I think. I also don't think it really matters if they go up to high numbers, the largest allowed version number is u64::MAX.

[ebfull]

Pushed 6d0edc8cd: end-of-support widened 7 → 18 days so the halt lands one week after the expected Ironwood activation (~2026-08-04, height ~3,436,436) instead of ~4 days before it; ESTIMATED_RELEASE_HEIGHT refreshed to 3,415,700 for a same-day cut and changelog dates moved to 2026-07-17. CHANGELOG_PARAMS.md has the re-tuning row; PR description updated to match.