* Fix IBD sync conditions
* Fix syntax
* Fix Sprintf
* Bump version
* On negotiation check only blocks in future of PP
* Only log error and add comment
* Fix comment
* Compare blue score with selected tip when checking if a pruning point proof is needed
* Don't redeclare err
Co-authored-by: Michael Sutton <mikisiton2@gmail.com>
* fix mempool accessing, rewrite get_mempool_entries_by_addresses
* fix counter, add verbose
* fmt
* addresses as string
* Define error in case utxoEntry is missing.
* fix error variable to string
* stop tests from failing (see in code comment)
* access both pools in the same state via parameters
* get rid of todo message
* fmt - very important!
* perf: scriptpublickey in mempool, no txscript.
* address reveiw
* fmt fix
* mixed up isorphan bool, pass tests now
* do map preallocation in mempoolbyaddresses
* no proallocation for orphanpool sending.
Co-authored-by: Ori Newman <orinewman1@gmail.com>
* Illustrate the bug through prints
* Change consensus API to a single ResolveVirtual call
* nil changeset is not expected when err=nil
* Fixes a deep bug in the resolve virtual process
* Be more defensive at resolving virtual when adding a block
* When finally resolved, set virtual parents properly
* Return nil changeset when nothing happened
* Make sure the block at the split point is reversed to new chain as well
* bump to version 0.12.4
* Avoid locking consensus twice in the common case of adding block with updateVirtual=true
* check err
* Parents must be picked first before set as virtual parents
* Keep the flag for tracking virtual state, since tip sorting perf is high with many tips
* Improve and clarify resolve virtual tests
* Addressing minor review comments
* Fixed a bug in the reported virtual changeset, and modified the test to verify it
* Addressing review comments
* Logger: change log level for "Couldn't find UTXO entry" to debug
* ignore error for orphans
* continue also if orphans
* check if blocks exist
* safe rpc mode
* limit window size
* verify block status depending on context
* allow a 2 factor gap in expected mergeset size
Co-authored-by: msutton <mikisiton2@gmail.com>
* Avoid creating the chain iterator if high hash is actually low hash
* Always use iterator in nextPruningPointAndCandidateByBlockHash
* Initial failing test
* Minimal failing test + some comments
* go lint
* Add simpler tests with two different errors
* Missed some error checks
* Minor
* A workaround patch for preventing the missing utxo child diff bug
* Make sure we fully resolve virtual
* Move ResolveVirtualWithMaxParam to test consensus
* Mark virtual not updated and loop in batches
* Refactor: remove VirtualChangeSet from functions return values
* Remove workaround comments
* If block has no body, virtual is still considered updated
* Remove special error ErrReverseUTXODiffsUTXODiffChildNotFound
Co-authored-by: Ori Newman <orinewman1@gmail.com>
* Move OnBlockAdded event to the channel that was only used by virtualChanged events
* Don't send notifications for header-only and invalid blocks
* Return block status from block processor and use it for event raising decision
* Use MaybeEnqueue for consensus events
* go lint
* Fix RPC call to actually include tx ids
Co-authored-by: msutton <mikisiton2@gmail.com>
* Use a channel from within consensus in order to raise change events in order -- note that this is only a draft commit for discussion
* Fix compilation
* Check for nil
* Allow nil virtualChangeChan
* Remove redundant comments
* Call notifyVirtualChange instead of notifyUTXOsChanged
* Remove redundant comment
* Add a separate function for initVirtualChangeHandler
* Remove redundant type
* Check for nil in the right place
* Fix integration test
* Add data to virtual changeset and cleanup block added event logic
* Renames
* Comment
Co-authored-by: Ori Newman <orinewman1@gmail.com>
* Block template cache improvement
* Avoid concurrent calls to template builder
* Clear cache on new block event
* Move IsNearlySynced logic to within consensus and cache it for each template
* Use a single consensus call for building template and checking synced
Co-authored-by: msutton <mikisiton2@gmail.com>
* Use separate than finality depth for merge set calculations after HF
* Add comments and edit error messages
* Fix TestValidateTransactionInContextAndPopulateFee
* Don't disconnect from node if isViolatingBoundedMergeDepth
* Use new merge root for virtual pick parents; apply HF1 daa score split for validation only
* Use `blue work` heuristic to skip irrelevant relay blocks
* Minor
* Make sure virtual's merge depth root is a real block
* For ghostdag data we always use the non-trusted data
* Fix TestBoundedMergeDepth and in IBD use VirtualMergeDepthRoot instead of MergeDepthRoot
* Update HF1DAAScore
* Make sure merge root and finality are called + avoid calculating virtual root twice
* Update block version to 1 after HF
* Update to v0.12.0
Co-authored-by: msutton <mikisiton2@gmail.com>
* protobuf for new block template notification structs
* appmessage and wire for new block template notification structs
* Set up the entire handler/call-chain for managing the new-block-template event
* Fix a bug in the case where syncer chain is fully known to syncee
* Extract chain negotiation to a separated method
* Bump version and update the changelog
* Add zoom-in progress validation and some debug logs
* Improved error explanation
* go fmt
* Validate zoom-in progress through a total count
* a patch for fixing p2p v4 IBD issues for all side-chains
* Perform side-chain check earlier to avoid IBD start
* A few comments explaining the IBD patch
* Improve output of non-critical protocol errors to avoid user panic
* Add log messages at the end of IBD with headers proof
* Found a case where this was falsely triggered due to the wrong equality test
* The first message is expected to contain headers and not a "done" message (+comment and error text fixes)
* Dequeue w/o timeout during pp anticone batch processing
* Add a verification step for catching possible new IBD errors
* Fetch missing bodies for both, syncer selected tip past and relay block past
* Make sure the syncer is behaving correctly to avoid out of index errors
* Make sure progress reporter does not exceed 100%
* No orphan roots, so no need to queue the empty list
* Add a log to report utxo fetch failure with err message
* A duplicate blocks should not appear as a warning
* typo
* add p2p v5 which is currently identical to v4
* set all internal imports to v5
* set default version to 5
* Send pruning point and its anticone in batches
* go lint
* Fix jsom format
* Use DequeueWithTimeout
* Assert that batch size < route capacity
* oops, this is a flow handler, by definition it needs to be w/o a timeout
* here however, a timeout is required
* Keep IDs of prev messages unmodified
* previous merge operation accidentally erased an important part of this pr
* Extend timeout of simple sync
Co-authored-by: Ori Newman <orinewman1@gmail.com>
* add p2p v5 which is currently identical to v4
* set all internal imports to v5
* wip
* set default version to 5
* protobuf gen for new ibd chain locator
* wire for new ibd chain locator types
* new ibd shared block algo -- only basic test passing
* address the case where pruning points disagree, now both IBD tests pass
* protobuf gen for new past diff request message
* wire for new request past diff message
* handle and flow for new request past diff message - logic unimplemented yet
* implement ibd sync past diff of relay and selected tip
* go fmt
* remove unused methods
* missed one err check
* addressing simple comments
* apply the traversal limit logic and sort headers
* rename pastdiff -> anticone
* apply Don't relay blocks in virtual anticone #1970 to v5
* go fmt
* Fixed minor comments
* Limit the number of chain negotiation restarts
* Report progress percentage when downloading headers in IBD.
* Extract reporting logic to a separate type.
* Report progress for IBD missing block bodies.
* Separate IBD to a new flow (so now invs are handled concurrently and no route capacity errors)
* Invs messages should be queued while waiting for BlockLocator msg
* Close IBD channel so that HandleIBDFlow exits too
* Apply flow separation to p2p protocol v4
* Manage the IBDRequestChannel through the Peer struct
* Some IBDs take a little longer
* Copy blockrelay flows to v4
* Remove duplicate sending of the same DAA blocks
* Advance testnet version
* Renames and add comments
* Add IBD test between v3 and v4
* Fix syncee v4 p2p version
* Check if netsync finished with selected tip
* Implement upgrade mechanism for p2p
* Remove dependencies from flowcontext to v3
* Add p2p v4
* Add Ready flow
* Remove copy paste code of v3
* Register SendAddresses flow at the top level
* Add option to set protocol version from CLI and add TestAddressExchangeV3V4
* Send ready message on minimal net adapter
* Rename defaultMaxProtocolVersion->maxAcceptableProtocolVersion
* Update to version v0.11.8
* Call OnPruningPointUTXOSetOverride after committing the staging consensus
* use domain.Consensus() in utxo index so it'll always check the right consensus
* Get rid of genesis's UTXO dump
* Allow known orphans when AllowSubmitBlockWhenNotSynced=true
* gofmt
* Avoid IBD without changing the pruning point when only genesis is available
* Add DisallowDirectBlocksOnTopOfGenesis=true for mainnet
* Remove any mention to nobanning to let stability tests run
* Rename ifGenesisSetUtxoSet to loadUTXODataForGenesis
Co-authored-by: Ori Newman <>
* UTXO dump of block 0fca37ca667c2d550a6c4416dad9717e50927128c424fa4edbebc436ab13aeef
* Activate HF immediately and change reward to 1000
* Change protocol version and datadir location
* Delete comments
* Fix zero hash to muhash zero hash in genesis utxo dump check
* Don't omit genesis as direct parent
* Fix tests
* Change subsidy to 500
* Dont assume genesis multiset is empty
* Fix BlockReward test
* Fix TestValidateAndInsertImportedPruningPoint test
* Fix pruning point genesis utxo set
* Fix tests related to mainnet utxo set
* Dont change the difficulty before you have a full window
* Fix TestBlockWindow tests
* Remove global utxo set variable, and persist mainnetnet utxo deserialization between runs
* Fix last tests
* Make peer banning opt-in
* small fix for a test
* Fix go lint
* Fix Ori's review comments
* Change DAA score of genesis to checkpoint DAA score and fix all tests
* Fix the BlockLevel bits counting
* Fix some tests and make them run a little faster
* Change datadir name back to kaspa-mainnet and change db path from /data to /datadir
* Last changes for the release and change the version to 0.11.5
Co-authored-by: Ori Newman <orinewman1@gmail.com>
Co-authored-by: Ori Newman <>
Co-authored-by: msutton <mikisiton2@gmail.com>
* Modify DefaultTimeout to 120 seconds
A temporary workaround for nodes having trouble to sync (currently the download of pruning point related data during IBD takes more than 30 seconds)
* Cache existence in reachability store
* Cache block level in the header
* Fix IBD indication on submit block
* Add hardForkOmitGenesisFromParentsDAAScore logic
* Fix NumThreads bug in the wallet
* Get rid of ParentsAtLevel header method
* Fix a bug in BuildPruningPointProof
* Increase race detector timeout
* Add cache to BuildPruningPointProof
* Add comments and temp comment out go vet
* Fix ParentsAtLevel
* Dont fill empty parents
* Change HardForkOmitGenesisFromParentsDAAScore in fast netsync test
* Add --allow-submit-block-when-not-synced in stability tests
* Fix TestPruning
* Return fast tests
* Fix off by one error on kaspawallet
* Fetch only one block with trusted data at a time
* Update fork DAA score
* Don't ban for unexpected message type
* Fix tests
Co-authored-by: Michael Sutton <mikisiton2@gmail.com>
Co-authored-by: Ori Newman <>
* Send pruning point anticone in topological order
Fix a UTXO pagination bug
Lengthen the stabilization time for the last DAA test
* Extend "sudden hash rate drop" test length to 45 minutes
Co-authored-by: Kaspa Profiler <>
* Check blue score before requesting a pruning proof
* BuildPruningPointProof should return empty proof if the pruning point is genesis
* Don't fail many-tips if kaspad exits ungracefully
