Integration tests have been added to determine the IBFT components
response to the reception of RoundChange messages.
Through the course of creating these tests the following changes were
made to the RoundChangeManager:
* Stores RoundChangeMessages in the order of arrival
* Discards subsequent RoundChange messages from the same node,
targeting a specific round.
* NewRoundMessageValidator ignores Round Number when comparing blocks
NewRoundMessageValidator previously ensured the block in the newest
PreparedCertificate was the same as that in the Proposal by comparing
block hashes.
However, the hash-function for received IBFT blocks includes round
number - thus the blocks will never be deemed 'equal'.
As such, the blocks must be compared using a hash function which
ignores the round number - i.e. the OnChain function.
To test the IBFT implementation at an integration level, an amount of "setup" is required.
This commit creates a framework of "n" 'external' nodes which are able to inject messages to the IbftController, while also exposing the messages received from the local node.
Thus integration tests are able to be written in terms of input events (including messages), expected (network) responses and changes in the blockchain state.
The RoundChangeManager was incorrectly creating its internal
RoundChangeMessageValidator with "quorum" number of Prepare messages
required - however, given the ProposalMessage in the Prepared-
Certificate represents a valid node.
The IbftHeightManager is responsible for all things 'meta-round'
related, such things as:
* Handling RoundTimeout and starting a new round
* Handling NewRound messages and starting a new round
* Ensuring RoundChange messages are sent at the correct time with
appropriate content.
* Collating RoundChange messages and starting a new round using the
best prepared certificate in the collection.
When the local node is to be the proposer for a new round
(due to round change), the new Ibft Round is to be created using the
latest received PreparedCertificate, and a NewRound message multicast
to all validators.
It should be noted, the block in the NewRound Proposal must be
modified from that received in the PreparedCertificate such that it
contains the executing round number.
This has required the MessageValidatorFactory used as part of the
RoundChangeMessageValidator to be renamed, which in turn has had
renaming knock on effects through the code.
An abstraction existed which allowed for the differentiation of IBFT
messages which were used 'in-round' (Proposal, Prepare, Commit) and
those that were inter-round (RoundChange, NewRound).
This abstraction has not proven useful now that actual capability has
been developed, and as such is being removed.
IbftBlockCreator is required to correctly set the MixHash, and also ensure the IbftBlockHashing function used to hash the block includes the round number, but not the commit seals.
* Upgrade errorprone
* Upgrade errorprone from 2.3.1 to 2.3.2
* Upgrade Jenkinsfile so that CI will use Java 11
* Suppress these new rules:
* EqualsGetClass
* ImmutableEnumChecker
* UnnecessaryParentheses
* Change code to conform to these new rules:
* BadImport
* BadInstanceof
* InconsistentHashCode
* LockNotBeforeTry
* MathAbsoluteRandom
* ModifiedButNotUsed
* UndefinedEquals
Received New Round messages require significant validation to ensure their content is consistent and valid with regard the current state of the system.
This commit includes a class (and tests) which ensures illegally constructed NewRound messages are identified.
This does not include the action to be taken upon detection of an illegal New Round message.
The classes used around the IBFT Messaging classes have all been renamed to make their use less cumbersome throughout the codebase.
Other issues are yet to be found (not all variable names and function names have been captured in this sweep).
A problem was identified whereby the vote in a child block was applied to the parent, this was resolved by
duplicating the parent VoteTally before modifying it, and injecting to the child block.
Some refactoring was also conducted to make VoteTally logic simpler.
Reverts #339
Görli's testnet regularly has proposers miss their turn and their
genesis block actually had a bad validator specified (that was voted out
as soon as they could) So many blocks do not have the proper "proposer"
as their block generator. As a consequence the
CliqueDifficultyValidationRule fails regularly, in fact it fails at
block 3, so this should be easy to re-create locally.
Clique Proposer Selection would choose an incorrect peer
when a signer was removed from the pool, as the algorithm worked
purely on the block count.
The algorithm has now been updated to ensure the next proposer is
incrementally the next signer in the ordered list, based on the
parent's proposer
RoundChangeValidator has been added which is responsible for
validating the content of a RoundChange message by using the
underlying MessageValidator capabilities.
Upgrade dependency versions
* All version information is merged into one place
* Upgrade most version to their latest non-test version.
* dependencies are now sorted
Two dependencies were not upgraded:
errorprone - There are new checks that require build or source changes.
vertex - The license to eclipse 2.0 which is not in our current approved license list.
The change is trivial but for tracability it should be done alone.
The ProposerSelection logic has been reworked such that it now accepts
a BlockInterface, through which IBFT parameters can be accessed
directly from the block header, without requiring literal hashing etc.
This has required the VoteBlockInterface to be renamed to
BlockInterface.
IBFT requires various aspects of the mining infrastructure in order
to create a proposed block.
This includes specifically the BlockCreator and MiningCoordinating,
the mining executor is not required at this stage, nor is the miner.
* Remove release/retain requirement from MessageData.
MessageData now wraps a BytesValue rather than a Netty ByteBuf so we no longer need to call release/retain and pass that through to the underlying ByteBuf.
Entirely removed support for a BytesValue wrapping a Netty ByteBuf as it does not provide any way to release the underlying ByteBuf.
* Add BytesValue.copyTo(byte[]) to avoid the need to call getArrayUnsafe() to copy the data.
Removing dependency upon using map.entry to define a vote that has
been proposed via JSON RPC.
This is part of the general clean up of the voting infrastructure
used for IBFT, Clique and legacy IBFT.
The consensus-mechanism specific block values have been separated
from the VoteTallyUpdater such that there is a single updater rather
than one per consensus mechanism.
This has necessitated the creation of a custom serialiser/
deserialiser for each mechanism instead.
This change will ultimatley bleed through to the proposed votes and
their insertion to mined blocks.
Update value returned by web3_clientVersion to be ethstats friendly
* Version part starts with a v
* SNAPSHOT builds report 32 bits of the git hash
* OS and architecture are sniffed out and reported
* JVM version and branding are sniffed out and reported
Example values (not all real):
pantheon/v0.9.0-dev-f800a0b1/osx-x86_64/oracle-java-1.8
pantheon/v0.9.0-dev-27960b57/osx-x86_64/zulu-java-11
pantheon/v0.9.0/linux-arm64/openjdk-java-12
Jason Frame
tmohay
Rob Dawson
Chris Mckay
Danno Ferrin
Roberto Saltini
Adrian Sutton