Release 2.0.10

This is a low-priority hotfix release to address two bugs in the block downloader. The chainstate directory of 2.0.10 is compatible with 2.0.9. If booting up a node from genesis, or an existing node has stalled in downloading blocks, this hotfix is necessary for your node.

Fixed

• Bug in microblocks inventory vector calculation that included invalidated microblocks as present bit. This bug will impact nodes booting up from genesis, but not affect nodes currently running at the chain tip (#2518).
• Bug in microblocks downloader logic that would cause the stacks-node to fail to wake-up to process newly arrived microblocks in certain instances (#2491).

Release 2.0.9

This is a low-priority hotfix release that patches a TOCTTOU bug in how the node handles blocks uploaded to it by 3rd parties, which if executed may lead to a denial-of-service (node crash) in the node while it is bootstrapping. People who run public nodes should upgrade to this release on their next node restart. Private nodes, such as those running behind a NAT or are otherwise unreachable from the Internet, are not affected.

Release 2.0.8

Upgrade priority: high for miners; medium for everyone else.

This hotfix repairs a denial-of-service vulnerability in the way the Stacks node processes smart contracts. A buggy or malicious smart contract transaction can cause the node to crash while processing it. Node operators should upgrade to this release as soon as possible.

If your node crashes due to this bug, you will need to restart with a new working directory. The denial-of-service bug, if executed, will corrupt the chainstate database. If you restart your node yourself, then you do not need to re-sync.

---

This is a hotfix release for improved handling of static analysis storage and
improved at-block behavior. The chainstate directory of 2.0.8 is compatible with
the 2.0.7 chainstate.

Fixed

• Improved static analysis storage
• at-block behavior in clarity-cli and unit tests (no changes in stacks-node
  behavior).

Release 2.0.7

This is an emergency hotfix that prevents the node from accidentally deleting
valid block data if its descendant microblock stream is invalid for some reason.

Fixed

• Do not delete a valid parent Stacks block.

Release 2.0.6

[2.0.6] - 2021-02-15

The database schema has not changed since 2.0.5, so when spinning up a
2.0.6 node from a 2.0.5 chainstate, you do not need to use a fresh
working directory. Earlier versions' chainstate directories are
incompatible, however.

Fixed

• Miner RBF logic has two "fallback" logic changes. First, if the RBF
  logic has increased fees by more than 50%, do not submit a new
  transaction. Second, fix the "same chainstate hash" fallback check.
• Winning block txid lookups in the SortitionDB have been corrected
  to use the txid during the lookup.
• The miner will no longer attempt to mine a new Stacks block if it receives a
  microblock in a discontinuous microblock stream.

Release 2.0.5

Note: This release contains a database schema change and therefore a fresh working directory must be used to spin up the node.

During this, miners are recommended to use the new --mine-at-height command line argument which will instruct the node to not mine until the Stacks chain has synchronized to the provided Stacks block height.

Added

• Miner heuristic for handling relatively large or computationally
  expensive transactions: such transactions will be dropped from the
  mempool to prevent miners from re-attempting them once they fail.
  Miners can also now continue processing transactions that are
  behind those transactions in the mempool "queue".

Fixed

• Miner block assembly now uses the correct block limit available via
  the node config
• tx_fees_streamed_produced fees are included in miner coinbase
  events for event observers
• SQLite indexes are now correctly created on database instantion

Changed

• STX unlock events are now sent over the events endpoint bundled
  into an associated unlock transaction
• Atlas attachments networking endpoints are disabled for this
  release, while networking issues are addressed in the
  implementation

Release 2.0.4

Merge pull request #2426 from blockstack/fix/disable-atlas

Disabling Atlas

Release 2.0.3

Note: This release contains a database schema change and therefore a fresh working directory must be used to spin up the node.

During this, miners are recommended to use the new --mine-at-height command line argument which will instruct the node to not mine until the Stacks chain has synchronized to the provided Stacks block height.

Added

• stacks-node --mine-at-height commandline option, which tells the
  stacks-node not to mine until it has synchronized to the given
  Stacks block height
• A new RPC endpoint /v2/blocks/upload/{consensus_hash} that accepts
  an uploaded Stacks block for a given sortition

Changed

• Enabled WAL mode for the chainstate databases. This allows much more
  concurrency in the stacks-node, and improves network performance
  across the board. NOTE: This changed the database schema, any
  running node would need to re-initialize their nodes from a new chain
  state when upgrading.
• Default value wait_time_for_microblocks: from 60s to 30s
• The mempool now performs more transfer semantics checks before admitting
  a transaction (e.g., reject if origin = recipient): see issue #2354
• Improved the performance of the code that handles GetBlocksInv p2p
  messages by an order of magnitude.
• Improved the performance of the block-downloader's block and
  microblock search code by a factor of 5x.

Fixed

• Miner mempool querying now works across short-lived forks: see issue #2389
• JSON deserialization for high-depth JSON objects
• Atlas attachment serving: see PR #2390
• Address issues #2379, #2356, #2347, #2346. The tracking of the
  LeaderBlockCommit operations inflight is improved, drastically
  reducing the number of block commit rejections. When
  aLeaderBlockCommit is not included in the Bitcoin block it was
  targeting, it is condemned to be rejected, per the Stacks
  consensus. To avoid wasting BTC, the miner now tries to send its
  next LeaderBlockCommit operations using the UTXOs of the previous
  transaction with a replacement by fee. The fee increase increments
  can be configured with the setting rbf_fee_increment.

Release 2.0.2

cargo fmt

Release 2.0.1

Merge pull request #2344 from blockstack/fix/peer-versions

Fix: peer version to 0x18 rather than 0x17