Zebra 2.5.0

Zebra 2.5.0 - 2025-08-07

This release includes the implementation of Zcash Network Upgrade 6.1 (NU6.1) on Testnet and sets its activation height on the public Testnet at block 3,536,500. Please update your Testnet nodes as soon as possible to ensure compatibility.

Breaking Changes

• Value pool "deferred" changes its identifier to "lockbox". getblock and getblockchaininfo RPC methods will now return lockbox as the FS_DEFERRED value pool to match zcashd. (#9684)

Added

• Implement one-time lockbox disbursement mechanism for NU6.1 (#9603, #9757, #9747, #9754)
• NU6.1 Testnet implementation and deployment (#9762, #9759)

Fixed

• validateaddress RPC bug (#9734)
• Count sigops using generics (#9670)

Contributors

Thank you to everyone who contributed to this release, we couldn't make Zebra without you:
@Galoretka, @arya2, @conradoplg, @dorianvp, @gustavovalverde, @oxarbitrage, @pacu and @upbqdn

Zebra 2.4.2

This release fixes a database upgrade bug that was introduced in the 2.4.0
release (which has been removed). If you have upgraded to 2.4.0, your Zebra
address index has become corrupted. This does not affect consensus, but will
make the RPC interface return invalid data for calls like getaddressutxos and
other address-related calls.

(Also refer to the 2.4.0 release notes below for important breaking changes.)

Zebra 2.4.2 prints a warning upon starting if you have been impacted by the bug.
The log line will look like:

2025-07-17T17:12:41.636549Z  WARN zebra_state::service::finalized_state::zebra_db: You have been impacted by the Zebra 2.4.0 address indexer corruption bug. If you rely on the data from the RPC interface, you will need to recover your database. Follow the instructions in the 2.4.2 release notes: https://github.com/ZcashFoundation/zebra/releases/tag/v2.4.2 If you just run the node for consensus and don't use data from the RPC interface, you can ignore this warning.

If you rely on the RPC data, you will need to restore your database. If you have
backed up the state up before upgrading to 2.4.0, you can simply restore the backup
and run 2.4.2 from it. If you have not, you have two options:

• Stop Zebra, delete the state (e.g. ~/.cache/zebra/state/v27/mainnet and
  testnet too if applicable), upgrade to 2.4.2 (if you haven't already), and
  start Zebra. It will sync from scratch, which will take around 48 hours to
  complete, depending on the machine specifications.
• Use the copy-state subcommand to regenerate a valid state.
  This will require an additional ~300 GB of free disk size. It is likely that
  it will take around the same time as syncing from scratch, but it has the
  advantage of not depending on the network.
  • Stop Zebra.
  • Rename the old corrupted state folder, e.g. mv ~/.cache/zebra ~/.cache/zebra.old
  • Copy your current zebrad.toml file to a zebrad-source.toml file and edit
    the cache_dir config to the renamed folder, e.g. cache_dir = '/home/zebrad/.cache/zebra.old'
  • The copy-state command that will be run requires a bigger amount of opened
    files. Increase the limit by running ulimit -n 2048; refer to your OS
    documentation if that does not work.
  • Run the copy-state command: zebrad -c zebrad-source.toml copy-state --target-config-path zebrad.toml. The command will take several hours to
    complete.

Fixed

• Fix 2.4.0 DB upgrade; add warning if impacted (#9709)
• Downgrade verbose mempool message (#9700)

Contributors

Thanks to @ebfull for reporting the bug and helping investigating its cause.

Zebra 2.4.1 - [REMOVED]

This version of Zebra wasn't fully published; it was tagged but the tag was
removed, and it was published on crates.io but it was yanked. It was not
published on Docker Hub.

We removed it due to a panic that happened during the pre-release validation.
However, we determined that the panic was caused by an external tool (ldb checkpoint) being used internally to make database backups and it was not a bug
in Zebra.

Zebra 2.4.0 - 2025-07-11 [REMOVED]

Breaking Changes

This release has the following breaking changes:

• This release contains a major database upgrade. It will upgrade an existing
  database, automatically moving it from the v26/ folder to a new v27/
  folder. However, downgrading is not possible. If you want to keep the
  possibility of reverting Zebra in case of an unexpected issue, backup the
  v26/ folder before running the upgraded Zebra. Note that the upgrade is slow
  and could take several hours or more to complete on smaller machines. Zebra will
  operate normally during that time; the only difference is that some RPC
  responses might return empty or not accurate data (pool values for arbitrary
  block heights and received balances for addresses).
• While this was never documented as an option for backups, if you relied on the
  ldb checkpoint tool to generate database backups, be advised that the tool
  is no longer supported and will corrupt databases generated or touched by
  Zebra 2.4.0 or later releases.
• The debug_like_zcashd config option for mining is no longer available. It
  was not enabled by default; if it is now present in the config file, Zebra
  will panic. Simply delete the config option to fix.
• The cm_u field byte order was fixed in getrawtransaction/getblock
  response, so if you relied on the wrong order, you will need to fix your
  application.
• The zebra-scan and zebra-grpc crates are no longer supported and were
  removed from the codebase.

Security

• Fix a consensus rule violation in V5 coinbase transactions at low heights. This issue could only occur on Regtest or custom testnets and is now resolved (#9620)

Added

• Implemented deserialization for Zebra's block and transaction types (#9522)
• Update getaddressbalance RPC to return received field (#9295, #9539)
• Added a mempool_change() gRPC method for listening to changes in the mempool (#9494)
• Added raw_value feature to serde_json (#9538)
• Modified zebra_network::Config type to use IPv6 listen_addr by default (#9609)
• Added invalidateblock and reconsiderblock RPC methods (#9551)
• Updated (z_)validateaddress to validate TEX addresses (#9483)
• Added a addnode RPC method (#9604)
• Added missing fields to getrawtransaction (#9636)
• Added value pool balances to getblock RPC output (#9432, #9539)
• Added support for configuring shielded addresses for mining (#9574)
• Added binding_sig, joinsplit_pub_key and joinsplit_sig fields to getrawtransaction/getblock response (#9652)
• Added a method in zebra-rpc to allow validating addresses (#9658)

Changed

• Allow Zebra crates to be compiled with alternative versions of their dependencies (#9484)
• Updated README with Arch build patch (#9513)
• Renamed and moved exports in zebra-rpc (#9568)
• Upgraded DB format to support new fields in RPC outputs (#9539)
• Moved GBT RPCs into the main RPC server (#9459)
• Added a Nu6_1 variant to NetworkUpgrade (#9526)
• Use zcash_script’s new Script trait (#8751)
• Removed debug_like_zcashd config option (#9627)
• Sync all chains in TrustedChainSync::sync, add NonFinalizedStateChange gRPC method (#9654)
• Added prometheus as a default feature in zebrad (#9677)

Fixed

• Preserve order of RPC output fields (#9474)
• Fixed cm_u field byte order in getrawtransaction/getblock response (#9667)
• Avoid repeatedly converting transactions to librustzcash types when computing sighashes (#9594)
• Correctly set optional scriptPubKey fields of transactions in getblock and getrawtransaction RPC outputs (#9536)
• Allow local outbound connections on Regtest (#9580)
• Allow for parsing z_gettreestate output type where optional fields are omitted (#9451)

Removed

• Removed zebra-scan and zebra-grpc (#9683)

Contributors

Thank you to everyone who contributed to this release, we couldn't make Zebra without you:
@ala-mode, @arya2, @conradoplg, @elijahhampton, @gustavovalverde, @idky137, @mpguerra, @oxarbitrage, @sellout, @str4d and @upbqdn

Zebra 2.3.0

Zebra 2.3.0 - 2025-05-06

Breaking Changes

• The RPC endpoint is no longer enabled by default in Docker. To enable it,
  follow the docs here.
• We will no longer be publishing Docker images tagged with the sha- or v
  prefixes. If you use tags starting with the v prefix, please update to
  images tagged N.N.N. For example, use 2.3.0 instead of v2.3.0. If you
  need a specific hash, each tag has a digest that you can use instead.
• All functionality that used to be guarded by the getblocktemplate-rpcs Cargo
  feature was moved out, and the feature is no longer present in the codebase.
  Note that all release builds following Zebra 1.3.0 had this feature enabled by
  default.

Added

• Track misbehaving peer connections and ban them past a threshold (#9201)
• Restore internal miner (#9311)
• Add reconsider_block method (#9260)
• Add NU7 constants (#9256)
• Add invalidate_block_method and invalidated_blocks field to non-finalized state (#9167)
• Add Transaction::V6 variants only (#9339)

Changed

• Downgrade verbose info message (#9448)
• Use read-only db instance when running tip-height or copy-state commands (#9359)
• Refactor format upgrades into trait (#9263)
• Remove the getblocktemplate-rpcs Cargo feature (#9401)
• Improve cache dir and database startup panics (#9441)

Fixed

• Remove a redundant startup warning (#9397)
• Advertise mined blocks (#9176)
• Ensure secondary rocksdb instance has caught up to the primary instance (#9346)
• Use network kind of TestnetKind in transparent addresses on Regtest (#9175)
• Fix redundant attributes on enum variants (#9309)

RPCs

• Add time and size fields to TransactionObject (#9458)
• Add inbound peers to getpeerinfo response (#9214)
• Extend getinfo (#9261)
• Add fields to getblockchaininfo RPC output (#9215)
• Add some missing fields to transaction object (#9329)
• Support negative heights in HashOrHeight (#9316)
• Add verbose support to getrawmempool (#9249)
• Fill size field in getblock with verbosity=2 (#9327)
• Add blockcommitments field to getblock output (#9217)
• Accept an unused second param in sendrawtransaction RPC (#9242)
• Make start and end fields optional and apply range rules to match zcashd (#9408)
• Return only the history tree root in GetBlockTemplateChainInfo response (#9444)
• Correctly map JSON-RPC to/from 2.0 (#9216)
• Permit JSON-RPC IDs to be non-strings (#9341)
• Match coinbase outputs order in Getblocktemplate (#9272)

Docker

• Refactor Dockerfile and entrypoint (#8923)
• Enhance Zebra configuration options and entrypoint logic (#9344)
• Better permission and cache dirs handling in Docker (#9323)
• Allow r/w access in mounted volumes (#9281)

Documentation

• Update examples for running Zebra in Docker (#9269)
• Add architectural decision records structure (#9310)
• Add Mempool Specification to Zebra Book (#9336)
• Complete the Treestate RFC documentation (#9340)

Contributors

@AloeareV, @Metalcape, @PaulLaux, @VolodymyrBg, @aphelionz, @arya2, @conradoplg,
@crStiv, @elijahhampton, @gustavovalverde, @mdqst, @natalieesk, @nuttycom,
@oxarbitrage, @podZzzzz, @sellout, @str4d, @upbqdn and @zeroprooff.

Zebra 2.2.0

Zebra 2.2.0 - 2025-02-05

In this release, Zebra introduced an additional consensus check on the branch ID of Nu6 transactions
(which is currently also checked elsewhere; but we believe it's important to check on its own to protect
against future code changes), along with important refactors and improvements.

Added

• An index to track spending transaction ids by spent outpoints and revealed nullifiers (#8895)
• Accessor methods to zebra-rpc request/response types (#9113)
• getblock RPC method now can return transaction details with verbosity=2 (#9083)
• Serialized NU5 blocks to test vectors (#9098)

Changed

• Migrated from deprecated jsonrpc_* crates to jsonrpsee (#9059, #9151)
• Optimized checks for coinbase transactions (#9126)
• Avoid re-verifying transactions in blocks if those transactions are in the mempool (#8951, #9118)
• Allow transactions spending coinbase outputs to have transparent outputs on Regtest (#9085)

Fixed

• Respond to getblockchaininfo with genesis block when empty state (#9138)
• Verify consensus branch ID in SIGHASH precomputation (#9139)
• More closely match zcashd RPC errors and getrawtransaction RPC behaviour (#9049)
• Fixes bugs in the lightwalletd integration tests (#9052)

Contributors

Thank you to everyone who contributed to this release, we couldn't make Zebra without you:
@Fallengirl, @arya2, @conradoplg, @elijahhampton, @futreall, @gustavovalverde, @idky137, @mpguerra, @oxarbitrage, @rex4539, @rootdiae, @sandakersmann and @upbqdn

Zebra 2.1.0

This release adds a check to verify that V5 transactions in the mempool have the correct consensus branch ID;
Zebra would previously accept those and return a transaction ID (indicating success) even though they would
be eventually rejected by the block consensus checks. Similarly, Zebra also now returns an error when trying
to submit transactions that would eventually fail some consensus checks (e.g. double spends) but would also
return a transaction ID indicating success. The release also bumps
Zebra's initial minimum protocol version such that this release of Zebra will always reject connections with peers advertising
a network protocol version below 170,120 on Mainnet and 170,110 on Testnet instead of accepting those connections until Zebra's
chain state reaches the NU6 activation height.
The getblock RPC method has been updated and now returns some additional information
such as the block height (even if you provide a block hash) and other fields as supported
by the getblockheader RPC call.

Breaking Changes

• Upgrade minimum protocol versions for all Zcash networks (#9058)

Added

• getblockheader RPC method (#8967)
• rust-toolchain.toml file (#8985)

Changed

• Updated getblock RPC to more closely match zcashd (#9006)
• Updated error messages to include inner error types (notably for the transaction verifier) (#9066)

Fixed

• Validate consensus branch ids of mempool transactions (#9063)
• Verify mempool transactions with unmined inputs if those inputs are in the mempool to support TEX transactions (#8857)
• Wait until transactions have been added to the mempool before returning success response from sendrawtransaction RPC (#9067)

Contributors

Thank you to everyone who contributed to this release, we couldn't make Zebra without you:
@arya2, @conradoplg, @cypherpepe, @gustavovalverde, @idky137, @oxarbitrage, @pinglanlu and @upbqdn

Zebra 2.0.1

Zebra 2.0.1 - 2024-10-30

• Zebra now supports NU6 on Mainnet. This patch release updates dependencies required for NU6. The 2.0.0 release was pointing to the incorrect dependencies and would panic on NU6 activation.

Breaking Changes

• The JSON RPC endpoint has cookie-based authentication enabled by default. If you rely on Zebra RPC, wou will need to adjust your config. The simplest change is to disable authentication by adding enable_cookie_auth = false to the [rpc] section of the Zebra config file; refer to the docs. (This was added in 2.0.0 but we include here again for clarity)

Changed

• Use ECC deps with activation height for NU6 (#8960)

Zebra 2.0.0 - 2024-10-25 - [YANKED]

This release was intended to support NU6 but was pointing to the wrong version of dependencies which would make Zebra panic at NU6 activation. Use 2.0.1 instead.

Breaking Changes

• The JSON RPC endpoint has cookie-based authentication enabled by default. If you rely on Zebra RPC, wou will need to adjust your config. The simplest change is to disable authentication by adding enable_cookie_auth = false to the [rpc] section of the Zebra config file; refer to the docs. (This was added in 2.0.0 but we included here again for clarity)

Added

• NU6-related documentation (#8949)
• A cookie-based authentication system for the JSON RPC endpoint (#8900, #8965)

Changed

• Set the activation height of NU6 for Mainnet and bump Zebra's current network protocol version (#8960)

Contributors

Thank you to everyone who contributed to this release, we couldn't make Zebra without you:
@arya2, @gustavovalverde, @oxarbitrage and @upbqdn.

Zebra 2.0.0-rc.0

Zebra 2.0.0-rc.0 - 2024-10-11

This version is a release candidate for the Zcash NU6 network upgrade on the Mainnet. While this version does not yet include the NU6 Mainnet activation height or current protocol version, all required functionality and tests are in place.

Please note that support for this release candidate is expected to conclude prior to the NU6 activation heights.

Security

• Added Docker Scout vulnerabilities scanning (#8871)

Added

• Added Regtest-only generate and stop RPC methods (#8849, #8839, #8863)
• Added fields to getmininginfo RPC method response (#8860)
• Copied the Python RPC test framework from zcashd into Zebra (#8866)

Changed

• Regtest halving interval to match zcashd and added a configurable halving interval for custom testnets (#8888, #8928)
• Updates post-NU6 Major Grants funding stream address on Mainnet (#8914)

Fixed

• Remove debugging output by default in Docker image (#8870)
• Fixes a typo in configuration file path of the docker-compose file (#8893)
• Return verification errors from sendrawtransaction RPC method (#8788)
• Respond to getheaders requests with a maximum of 160 block headers (#8913)
• Avoids panicking during contextual validation when a parent block is missing (#8883)
• Write database format version to disk atomically to avoid a rare panic (#8795)

Contributors

Thank you to everyone who contributed to this release, we couldn't make Zebra without you:
@arya2, @dismad, @gustavovalverde, @oxarbitrage, @skyl and @upbqdn

Zebra 1.9.0

This release includes deployment of NU6 on Testnet, configurable funding streams on custom Testnets, and updates Zebra's end-of-support (EoS) from 16 weeks to 10 weeks so that it will panic before the expected activation height of NU6 on Mainnet.

It also replaces the shielded-scan compilation feature with a new zebra-scanner binary, adds a TrustedChainSync module for replicating Zebra's best chain state, and a gRPC server in zebra-rpc as steps towards zcashd deprecation.

Recovering after finalizing a block from a chain fork

Zebra doesn't enforce an end-of-support height on Testnet, and previous versions of Zebra could mine or follow a chain fork that does not
activate NU6 on Testnet at height 2976000. Once a block from a fork is finalized in Zebra's state cache, updating to a version of Zebra that
does expect NU6 activation at that height will result in Zebra getting stuck, as it cannot rollback its finalized state. This can be resolved
by syncing Zebra from scratch, or by using the copy-state command to create a new state cache up to height 2975999. To use the copy-state
command, first make a copy Zebra's Testnet configuration with a different cache directory path, for example, if Zebra's configuration is at the
default path, by running cp ~/.config/zebrad.toml ./zebrad-copy-target.toml, then opening the new configuration file and editing the
cache_dir path in the state section. Once there's a copy of Zebra's configuration with the new state cache directory path, run:
zebrad copy-state --target-config-path "./zebrad-copy-target.toml" --max-source-height "2975999", and then update the original
Zebra configuration to use the new state cache directory.

Added

• A zebra-scanner binary replacing the shielded-scan compilation feature in zebrad (#8608)
• Adds a TrustedChainSync module for keeping up with Zebra's non-finalized best chain from a separate process (#8596)
• Add a tonic server in zebra-rpc with a chain_tip_change() method that notifies clients when Zebra's best chain tip changes (#8674)
• NU6 network upgrade variant, minimum protocol version, and Testnet activation height (#8693, 8733, #8804)
• Configurable NU6 activation height on Regtest (#8700)
• Configurable Testnet funding streams (#8718)
• Post-NU6 funding streams, including a lockbox funding stream (#8694)
• Add value pool balances to getblockchaininfo RPC method response (#8769)
• Add NU6 lockbox funding stream information to getblocksubsidy RPC method response (#8742)

Changed

• Reduce the end-of-support halt time from 16 weeks to 10 weeks (#8734)
• Bump the current protocol version from 170100 to 170110 (#8804)
• Track the balance of the deferred chain value pool (#8732, #8729)
• Require that coinbase transactions balance exactly after NU6 activation (#8727)
• Support in-place disk format upgrades for major database version bumps (#8748)

Fixed

• Return full network upgrade activation list in getblockchaininfo method (#8699)
• Update documentation for the new zebra-scanner binary (#8675)
• Update documentation for using Zebra with lightwalletd (#8714)
• Reduce debug output for Network on Regtest and default Testnet (#8760)

Contributors

Thank you to everyone who contributed to this release, we couldn't make Zebra without you:
@arya2, @conradoplg, @oxarbitrage, @therealyingtong and @upbqdn

Zebra 1.8.0

Zebra 1.8.0 - 2024-07-02

• Zebra now uses a default unpaid actions limit of 0, dropping transactions with
  any unpaid actions from the mempool and excluding them when selecting
  transactions from the mempool during block template construction.
• The zebrad binary no longer contains the scanner of shielded transactions.
  This means zebrad no longer contains users' viewing keys.
• Support for custom Testnets and Regtest is greatly enhanced.
• Windows is now back in the second tier of supported platforms.
• The end-of-support time interval is set to match zcashd's 16 weeks.
• The RPC serialization of empty treestates matches zcashd.

Added

• Add an init function for a standalone ReadStateService (#8595)
• Allow configuring more parameters on custom Testnets and an NU5 activation height on Regtest (#8477, #8518, #8524, #8528, #8636)
• Add default constructions for several RPC method responses (#8616, #8505)
• Support constructing Canopy proposal blocks from block templates (#8505)

Docs

• Document custom Testnets, Regtest and how they compare to Mainnet and the default Testnet in the Zebra book (#8526, #8636)

Changed

• Use a default unpaid action limit of 0 so that transactions with unpaid actions are excluded from the mempool and block templates by default (#8638)
• Lower the mandatory checkpoint height from immediately above the ZIP-212 grace period to immediately below Canopy activation (#8629)
• Use the new zcash_script callback API in zebra-script, allowing Zebra's ECC dependencies to be upgraded independently of zcash_script and zcashd (#8566)
• Put Windows in Tier 2 of supported platforms (#8637)
• Remove experimental support for starting the zebra-scan in the zebrad process as a step towards moving the scanner to its own process (#8594)
• Reduce the end of support time from 20 weeks to 16 weeks (#8530)
• Restore parts of the experimental internal-miner feature for use on Regtest (#8506)

Fixed

• Allow square brackets in network config listen addr and external addr without explicitly configuring a port in the listen addr (#8504)
• Fix general conditional compilation attributes (#8602)
• Update median_timespan() method to align with zcashd implementation (#8491)
  upbqdn marked this conversation as resolved.
• Allow custom Testnets to make peer connections (#8528)
• Refactor and simplify the serialization of empty treestates to fix an incorrect serialization of empty note commitment trees for RPCs (#8533)
• Fix port conflict issue in some tests and re-enable those tests on Windows where those issues were especially problematic (#8624)
• Fix a bug with trailing characters in the OpenAPI spec method descriptions (#8597)

Contributors

Thank you to everyone who contributed to this release, we couldn't make Zebra without you:
@arya2, @conradoplg, @gustavovalverde, @oxarbitrage and @upbqdn

Zebra 1.7.0

Zebra 1.7.0 - 2024-05-07

In this release we introduce Regtest functionality to Zebra and restored Windows support. Also adjusted our Zebra release interval from 2 weeks to 6 weeks approximately.

Added

• Preparing for upstream zcash_client_backend API changes (#8425)
• Regtest support (#8383, #8421, #8368, #8413, #8474, #8475)
• Allow Zebra users to contribute to the P2P network even if behind NAT or firewall (#8488)

Changed

• Adjust estimated release interval to once every 6 weeks and the end of support from 16 to 20 weeks (#8429)

Fixed

• Bump zcash script v0.1.15 and restore Windows support (#8393)
• Avoid possibly returning data from different blocks in z_get_treestate RPC method (#8460)
• Zebra panics with all features and no elasticsearch server available (#8409)

Contributors

Thank you to everyone who contributed to this release, we couldn't make Zebra without you:
@arya2, @oxarbitrage and @upbqdn