zakura-core/zakura v1.5.0

on GitHub

Added

• Added the ZIP 218 per-block shielded action limits at NU7. From the configured
  NU7 activation height onwards, blocks may contain at most 330 Orchard actions,
  330 Ironwood actions, 300 Sapling spends plus outputs, no Sprout JoinSplits,
  and a global shielded cost of 330 where each JoinSplit counts twice.
  getblocktemplate only selects transactions that fit these limits. Networks
  without an NU7 activation height follow today's consensus
  (#1027,
  #1041).
• Rejected mempool transactions whose own shielded actions exceed a ZIP 218
  per-block limit, with misbehavior score 100, because no block can include
  them. ZIP 218 does not specify this rejection
  (#1027).
• Set the ZIP 218 Sprout JoinSplit limit to zero instead of the 25 the ZIP
  names. ZIP 2003 disallows version 4 transactions at NU7, and only version 2,
  3, and 4 transactions can contain JoinSplits, so no block at or after NU7 can
  contain one (#1027).
• Recycle 60% of aggregate transaction fees into NSM from NU7 activation in
  every build. Round the contribution down once per block and require the
  coinbase to claim the remaining fees. Mining templates use the
  same split, and their coinbase fee reports only the fees collected by the
  miner (#1056).
• Added ZIP 218 "25-second Block Target Spacing" rules at NU7. From the
  configured NU7 activation height onwards, the rules apply a 25-second block
  target spacing, a 102-block difficulty averaging window, and a block subsidy
  divided by the spacing ratio. Networks without an NU7 activation height keep
  their existing consensus rules (#1066,
  superseding #851).
• Seeded the NSM balance before NU7 and rejected negative balances from NU7,
  with matching database migration and configurable initial balances
  (#1070).
• Verified reissuance rewards using the actual parent's NSM balance, with
  matching checkpoint funding, rollback, replay, and migration checks
  (#1073).
• Used parent NSM balances in mining templates, coinbase rewards, and subsidy
  RPCs after reissuance activation
  (#1074).
• Added network.zakura.nat_traversal to opt in to QUIC NAT traversal with
  native peers. It remains disabled by default; enabling it allows candidate
  address exchange and UDP probes but does not enable relays or external
  discovery (#968).
• Added block download and commit byte counters and apply queue depth to the
  existing continuous-sync samples for Dual and Zakura-only runs. The retained
  data can be used to generate charts
  (#984).
• getblockchaininfo now reports the ZIP 234 NSM value balance as
  nsmValueBalanceZat, so the counter can be observed and monitored. It is
  reported beside the monetary totals rather than inside them, so summing
  valuePools still yields the money supply
  (#1106).

Changed

• Extended v1.5.0 Mainnet support through approximately October 25, with the
  last supported height at 3,528,681
  (#1121).
• Updated the common libraries to 1.3.0-alpha.1, including the removal of
  unused experimental ZIP 233 fields
  (#1042).
• Retain 112 predecessor headers below the finalized anchor and backfill older
  validation contexts at startup, preparing storage for a wider difficulty window
  while preserving current consensus rules
  (#1045).
• Track signed issuance deficits from NU7 and migrate existing databases to format 29. Backfill only NU7 activation onward, leaving historical records unchanged when NU7 is unscheduled or the tip precedes activation. The counter excludes pre-NU7 funds and monetary totals. Older binaries require a pre-upgrade cache or a separate sync. (#1052).
• Every build rejects version 4 transactions from the NU7 activation height
  (ZIP 2003). Mainnet and default Testnet remain unscheduled;
  #1059 tracks their
  activation heights (#1060).
• Set the NU7 peer protocol versions to the ZIP 204 values: 170180 on Testnet
  and Regtest, and 170190 on Mainnet
  (#998,
  #1060).
• Scaled target-spacing-dependent node timing at NU7, including Testnet mining,
  RPC work estimation, sync and gossip lookahead, near-tip detection, stall
  warnings, and checkpoint progress hints
  (#1066).
• Reduced block commit latency by checking retained non-finalized headers in
  memory instead of reading each header from disk
  (#1069).
• Reduced block commit latency by skipping unused non-finalized UTXO snapshots
  for blocks without transparent prevout spends
  (#1076).
• Derive the initial NSM balance from local chain state without RPC, and check
  public-network seeds against the measured values. Configured networks derive
  their seed unless an explicit override is supplied, including zero. Reject
  configured schedules whose derived seed cannot fit the stored balance, and
  repair the seed row when an interrupted migration resumes under new settings
  (#1077).
• Derive NSM reissuance activation from the configured NU7 height using the
  spacing-aware third-halving crossover. Leave reissuance inactive when NU7 is
  unset or no qualifying crossover exists
  (#1082).
• Header-chain startup treats network-policy digest changes as diagnostic.
  Startup audits stored data under the configured policy, warns, and updates
  the digest automatically after a successful audit
  (#1085).
• Upgraded native P2P to Iroh 1.1 through the published Zakura compatibility
  crates and noq 1.2 while retaining the existing
  published Zcash cryptographic dependencies and network defaults. Native peers
  now require protocol 2; mixed old/new dual-stack peers retain legacy TCP, while
  native-only peers require the same transport cohort
  (#935).
• A configured native UDP port that is already occupied now fails startup
  instead of silently binding another port. Native endpoints bind only the
  configured address family, and automatic router port mapping is disabled
  (#935).
• Keep the connection and unrelated services running when a persistent native
  P2P stream write times out
  (#943).
• Trusted secondary nodes carry the primary's block receipt order and reset it
  when the primary process changes, preparing them for first-received mining.
  A primary restart temporarily lowers the secondary's reported height to its
  finalized tip while it reloads recent blocks
  (#1102).
• Bump zakura-state to 9.0.0 for the dedicated checkpoint handoff request and
  response variants. Consumers with exhaustive state request or response matches
  must handle the new variants
  (#972).
• Reduced header transition work by limiting checkpoint checks to affected
  heights
  (#1018).
• Reduced checkpoint finalization work when updating the retained header height
  (#1019).
• Reduced sync header processing work when the selected chain is unchanged
  (#1030).
• Reduced block commit latency by validating transparent spends against
  borrowed non-finalized UTXO indexes instead of cloning the complete UTXO set
  (#1109).
• Stream declares each stream's payload limits, message types, queue depths,
  and write policy as fields, replacing four Service hooks
  (#1115).
• Updated the Zakura Common (zakura-core/common) crates from 1.3.0-alpha.1
  to 2.0.0, replacing the upstream zcash_protocol, zcash_address, and
  zcash_transparent dependencies with the zakura-protocol, zakura-address,
  and zakura-transparent forks
  (#1144).
• Set the NU7 consensus branch ID to 0x77190AD9 from ZIP 259, replacing the
  test-only placeholder; NU7 remains unscheduled on Mainnet and Testnet
  (#1144).

Removed

• Removed the temporary nu7 / nu7-experimental build feature from
  zakura-consensus and zakurad. NU7 rules activate by height in every build
  (#958,
  #997,
  #1060).
• Remove the explicit nsm_reissuance_height configuration override; networks
  with NU7 configured use the calculated crossover
  (#1082).
• Removed the state.contextual.unspent_utxo_snapshot.duration_seconds and
  state.contextual.mined.unspent_utxo_snapshot.duration_seconds metrics
  because transparent spend validation no longer creates UTXO snapshots
  (#1109).

Fixed

• Fixed funding-stream address-period calculations on configured Testnets whose
  funding-stream range begins before the first address-period anchor
  (#1007).
• Fixed fork recovery rejecting replacement blocks using size estimates from
  another branch (#1020).
• Avoided generating an extra shielded coinbase proof while assembling block
  templates, reducing getblocktemplate latency for shielded mining addresses
  (#1035).
• Fixed the end-of-support halt and getdeprecationinfo estimate to follow
  target-spacing changes, preserving the configured support duration across
  network upgrades
  (#1043).
• Correct halving-height estimates on custom networks where a halving occurs
  before Blossom, counting the slow-start shift once
  (#1046).
• Reject non-positive custom halving intervals and intervals that overflow when
  converted to target seconds, preventing panics or incorrect halving indices
  (#1046).
• Use the candidate height for the Testnet mining-template minimum-difficulty
  heuristic and derive node timing estimates and limits from the target-spacing
  schedule, preserving the current NU7 spacing and averaging window
  (#1047).
• Preserved the existing 28-header difficulty-context API while allowing
  callers to supply the wider retained context needed for NU7
  (#1048).
• Preserve deferred funding during slow-start rollback and legacy database replay on configured
  networks. (#1049).
• Limit format-29 Deferred checks to balance changes from the NU7 baseline onward, rejecting inconsistent replay/commit histories without auditing the absolute historical balance. Skip all backfill work when NU7 is unscheduled or has not activated, avoiding unnecessary startup scans and resync requirements on configured networks. (#1052).
• Mark wider retained header validation contexts with a new database format and
  report finalized-state divergence before attempting context repair at startup
  (#1062).
• Fixed Regtest to derive its first halving height from its target spacings, so
  funding-stream recipient rotation remains aligned with subsidy halvings after
  NU7 (#1066).
• Fixed network-height estimates and near-tip detection windows across block
  target spacing changes
  (#1068).
• Bounded the NSM subsidy parent lookup, preserved negative-balance errors, and
  moved mining template proof construction off RPC workers with a shared
  concurrency limit, including long-poll and recovery templates
  (#1074).
• Kept empty templates with precomputed coinbases out of the proof queue so tip
  changes can return work while another proof build runs
  (#1087).
• getblocktemplate rebuilds a template whose parent changed while it was being
  built (a new block, an equal-height reorg, or a concurrent caller selecting the
  new tip first) instead of returning a template parent changed; retry error
  that made miners drop their current job
  (#1088).
• getblocktemplate now rebuilds a superseded template in its fallback-recovery
  path instead of returning a transient error, so miners keep receiving work when
  the chain tip moves during proposal validation
  (#1090).
• Wake idle internal mining solvers when new work arrives instead of waiting out
  a 20-second sleep
  (#1094).
• Mainnet mempool activation and crawling require both near-tip sync throughput
  and a local-clock estimate that the chain tip is within 100 blocks of the
  network tip. The estimate adds a freshness check when peer starvation produces
  zero-length sync rounds. An active mempool disables when the estimate exceeds
  the 1000-block rollback window. The mempool only scores transaction verification
  failures against peers while the estimate is within 100 blocks and verification
  used the current tip height. Branch-ID and lock-time mismatches no longer
  penalize mempool peers. Test networks bypass the clock estimate so sparse mining
  does not disable their mempools. The crawler yields while waiting for a fresh
  tip. Subscribers receive invalidation events when the mempool disables
  (#27).
• Prevented continuous-sync canaries from treating a valid Mainnet block gap
  as a stalled node by requiring an exact local header backlog before applying
  the stall deadline
  (#846).
• Reported continuous metrics failures as unavailable status evidence instead
  of misclassifying them as sync stalls
  (#846).
• Stopped a dual-stack run as soon as legacy fallback takes over block sync,
  and named the handoff as the failure reason instead of reporting missing
  header evidence
  (#846).
• Required exact committed heights from both nodes before the cluster monitor
  treats peer advancement as sync-stall evidence
  (#846).
• Retained native per-IP admission limits across concurrent connections to the
  same identity, and disabled the new QUIC NAT traversal address exchange and
  peer-directed probes
  (#935).
• Dual-stack checkpoint sync can recompute from current commitment trees when
  compatible native peers cannot supply VCT metadata. Nodes already using frozen
  VCT trees still wait for authenticated metadata
  (#935).
• Reduced header-fetch waits during integrated chain sync by preparing a
  checkpoint-sized prefix when admitted block work is running low
  (#938).
• Prevent checkpoint verification stalls when the verifier rejects a side-chain
  block while the state commits a verified checkpoint range, or when a canceled
  caller drops recovery from a failed state commit
  (#939).
• Prevent sync restart loops after transient UTXO lookup and post-checkpoint
  verification timeouts by retrying affected blocks while parent commits finish.
  Wait up to two minutes for pending commits before accepting duplicate blocks,
  reject forged checkpoint-handoff heights, and avoid overflow with high custom
  checkpoints. Report bodies that fail their transaction Merkle root as payload
  mismatches before checking the coinbase height. Reject and score missing inputs
  proven against any committed parent, before and after the UTXO wait. Drop
  blocks whose parent is neither committed nor in flight without restarting sync
  (#940).
• Keep native P2P session setup and recovery responsive under congestion, and
  release stream capacity safely when sessions end
  (#943).
• Reject disallowed messages and malformed buffered responses even after a
  session fails, while preserving block download stall penalties without
  penalizing local delays
  (#943).
• Settle outgoing GetBlocks request ownership exactly once across expiry,
  cancellation, and reset, including requests still queued for transport.
  Return every unsent height immediately when any part of a queued request
  expires, preserving received bodies and replacement requests. Discard
  already-committed heights during request cleanup after a forward reset.
  Retire requests skipped or expired before writing without charging the peer
  for an unanswered probe or timeout, including floor watchdog avoidance.
  Retire queued requests immediately when a competing body makes them unwritable,
  preserving the received body and returning the remaining unsent work.
  Give initial probes and queued responses time to arrive from slow peers,
  while preserving the shorter floor rescue deadline for measured peers in
  both byte-count and block-count modes
  (#944).
• Preserve the first invalid-block verdict when another invalid verdict arrives
  with different evidence or a different rule
  (#948).
• Refuse to resume an interrupted VCT sync on the legacy P2P stack, with a
  diagnostic identifying the required configuration
  (#948).
• Allow read-only access to an interrupted VCT database when checkpoint sync,
  VCT fast sync, or native P2P is disabled
  (#948).
• Return an error from getblocktemplate when the state cannot supply a complete
  difficulty window
  (#948).
• Fixed pruned nodes advertising block bodies they no longer retain to native
  block-sync peers, including after restarts and further pruning
  (#966).
• Fixed native sync stalling after its only VCT auxiliary supplier temporarily
  replies busy. Retry that supplier after its next valid status, even when its tip
  is unchanged, while keeping other suppliers available. Clear temporary retry
  history on disconnect or session replacement so a reconnected supplier can be
  used immediately. Rotate suppliers after successful repairs and refusals so each
  waiting peer gets a fair turn.
  Keep serving unchanged header paths while the node syncs, with coherent page
  snapshots and finalized metadata matched to the requested branch. Resolve
  ancestry outside commit locks and stop at the requested locator so small header
  requests do not stall state transitions while scanning the retained graph.
  Peer serving leases no longer prevent old forks from being pruned to admit a
  better-work branch. Release serving capacity before returning each page so idle
  peers cannot keep all general serving slots between requests. Reuse bounded
  continuation indexes so small pages do not repeat ancestry lookups
  (#970).
• Fixed incomplete committed-payload reporting after block-sync queue cleanup.
  Continuous-sync runs enable accurate accounting through the opt-in
  sync-metrics feature, with stable sampling windows for detailed commit rates.
  Native continuous-sync runs retain progress samples every 10 seconds.
  Normal builds omit the commit-byte counter
  (#986).
• Fixed a panic when a build without an NU7 consensus branch ID adds an NU7
  activation block to the history tree. The block now fails to commit with
  HistoryTreeError::MissingBranchId. VCT verification preserves this local
  error without rejecting peer deliveries. Startup rejects stored snapshots
  that require a missing branch ID before exposing the database. The format
  check also reports snapshot decoding errors. Fallible primitive history tree constructors
  reject upgrades without a branch ID
  (#998).
• Fixed getblockchaininfo and other ZEC-valued RPC responses failing to parse
  with "loss of precision parsing ZEC value", which affected roughly one in
  seven value pool balances and became constant after NU7
  (#1105).
• Keep sync moving after the final checkpoint when full-verification blocks are
  already queued and no further state requests arrive, without reporting a
  committed block as failed if the handoff notification is delayed
  (#972).
• Prevent semantic children from ending checkpoint sync before the configured
  boundary and disabling further checkpoint writes
  (#972).
• Reject Regtest configurations with insufficient mandatory checkpoints during
  configuration parsing instead of panicking during startup or block validation
  (#972).
• Fixed block-sync pauses when provider latency estimates expire, so ready blocks
  are requested promptly (#1006).
• Keep newly accepted forks available for extension at the retention limit and
  allow evicted blocks to be downloaded again. Clear stale header verification
  records after normal commits and reconsideration. Preserve shared transaction
  outputs when removing a fork so pending block verification can continue
  (#1104).

This release also ships Mainnet release state up to height 3,490,665.