← All calls
ARBORIST#45

Zebra mining success and zcashd 5.4.0

Testnet blocks were successfully mined with Zebra, and zcashd 5.4.0 exposed the total supply and the transparent pool balance.

Agenda summary 9

  1. Zebra rc.4 - testnet mining success

    Zcash Foundation · Pili

    • The team manually tested whether blocks can be mined with Zebra on the testnet. It mined several blocks using a modified s-nomp and a CPU miner, and plans to write up and publish the method.
    • Before asking mining pools and others to test on mainnet, the team plans to implement more RPCs that appear necessary. Once the remaining RPCs and performance improvements are done, it will look for test volunteers.
    • The fifth release candidate, Zebra 1.0.0-rc.4, was tagged, and rc.5 will come once the mining pool-related RPCs and performance improvements are finished.
  2. FROST update - reddsa integration and aggregation optimization

    Zcash Foundation · Deirdre

    • The team is removing duplication among signature implementations scattered across the FROST, redjubjub, reddsa and Zebra libraries. In the latest Zebra it removed the redpallas implementation that had been carried over from redjubjub's old FROST implementation.
    • Zebra now points to reddsa (0.40), and redjubjub 0.60 is a thin wrapper around reddsa.
    • The PR to put the FROST implementations of redjubjub and redpallas in reddsa, based on the latest spec-conformant FROST and its re-randomized variant, is being finished.
    • An aggregation optimization was merged into the FROST crate. Instead of checking every signature share, it builds the signature and verifies it with the group key, and only on failure verifies shares one by one to find the cheater. When everyone is honest, computation drops significantly.
    • Natalie, a new member of the Foundation, merged their first code into the FROST library last week.
  3. Zcashd 5.4.0 release

    Electric Coin Company · Kris, Str4d, Daira

    • 5.4.0 was tagged the day before. getblockchaininfo on a reindexed node provides the total supply and the size of the transparent pool, and a reindex is required to see this.
    • Bugs reported since 5.0.0 and the getblocktemplate backport for miners were also included.
    • Previously deprecated features such as getnewaddress, getrawchangeaddress, z_getnewaddress and z_listaddresses are now off by default.
    • The team recommends moving to the mnemonic seed derivation of 4.7.0 and the ZIP 32-based approach, and temporary use remains possible with the allow deprecated flag.
  4. Zcashd 5.4.0 - reindex issue and the 5.5.0 plan

    Electric Coin Company · Kris, Str4d, Daira

    • The team fixed a problem where Ctrl+C did not exit during reindex or rescan, and a bug where an Orchard wallet hit an assertion failure with a combination of reindex and rescan. It was not a fund-loss bug.
    • A known issue is that RPC locks up during reindex on some systems. Reindex is usually over 24 hours and can grow by 8 to 12 hours; it is fixed on master, so the team said to postpone reindexing until 5.5.0 if possible.
    • 5.5.0 centers on ZIP 317 support in both the wallet and the block template.
  5. librustzcash refactor and retroactive application of ZIP 216

    Electric Coin Company · Kris, Str4d, Daira, Nate

    • Str4d refactored the Sapling module to have an API similar to Orchard's. The aim is to update the transaction construction code so it does not block Zingolabs' Orchard work.
    • In the next release, ZIP 216 will be applied retroactively to forbid parsing non-canonical jubjub points. The Rust code has already changed, and the team confirmed that the pre-NU5 testnet and mainnet chains are not affected.
    • The impact is practically nil. It does not apply to users who generated keys by the ZIP 32 or ZIP 316 methods, and a sender cannot tamper with an address to send such notes. The concern was that, under the consensus rules, a sender could have created them.
    • If someone is affected, the symptom would be a wallet no longer being able to decrypt notes it previously decrypted. No intentional fix is planned, and the team asked people to report it if they experience it.
  6. ZIP 317 block construction and zero-conf chains

    Electric Coin Company · Daira, Str4d

    • Daira said the work of implementing ZIP 317 in the block construction algorithm is going well.
    • zcashd, like Bitcoin Core, allows chains of transparent zero-conf transactions in the mempool, so it needs to differ slightly from ZIP 317.
    • The approach is to hold back children before their parents are selected, and Daira plans to add the method to ZIP 317 as a note.
    • Str4d thought there would be no impact because Zebra does not keep zero-conf chains in its mempool.
  7. ZSA update - transfer implementation PR

    QEDIT · Vivek, Adi, Daira

    • The PR implementing transfers, milestone 5 of the ZSA project, was submitted. There are related PRs in orchard (PR 372), librustzcash and the test vectors.
    • The team had a preliminary discussion of backporting ZSA's cryptographic changes to the orchard protocol first, before the ZSA update. The idea is to merge incrementally so that large code does not land all at once.
    • Regular meetings with the ZIP editors, the Foundation and ECC began, and the team said design agreement is getting faster. The signature method change Adi asked about is a base point domain separation matter in the signature scheme, and Daira said it is not a problem.
  8. PoS research status and the Cosmos community call

    Electric Coin Company · Nate, Deirdre

    • Nate said there is almost no progress on PoS research, and that they want to put out a straw proposal for the transition plan in the first half of this year to get feedback. All engineers are focused on wallet performance.
    • The Foundation held a community call the previous day with Christopher Goes of the Tendermint and Cosmos ecosystem, discussing the BFT consensus layer and uses of staking. The video will be posted on YouTube.
    • Deirdre said the authenticated channels of Tendermint and Cosmos could make FROST signing practically free. Distributed key generation is harder, and trusted keygen is also possible.
  9. Open discussion - mempool limits and ZIP 401

    Adi, Str4d, Daira

    • Adi asked about the size limit of a node's mempool. Str4d said it uses the weighted random eviction described in ZIP 401. When the weight exceeds a threshold, transactions are evicted at random with weights proportional to the fee paid.
    • This method prevents an attacker from pushing out other transactions with specific pathological transactions. For an attacker to occupy x% of the mempool, they must create and maintain x% of the transactions. The fee-related parts are updated in ZIP 317.
    • Daira said zcashd differs slightly from the spec, for example by using estimated memory size rather than serialized size, but the impact is small.
    • An evicted transaction is blocked from being re-added until the next block, and usage can be seen through getmempoolinfo and Prometheus metrics.

This summary is based on the original meeting notes; check them for exact details.