I made an suboptimal (aka stupid) decision to stop and wait for peers
one by one which makes shutdown very slow - O(n). This PR decouples sending
stop signal from waiting a thread to exit. On top of it in Peers we
first send stop signal to all peers and only after that start waiting
for them to exit. It gives us a constant time of shutdown in most of the
cases.
* connection no longer wrapped in an Option in peer
* introduce peer.send()
* remove some Arc indirection
* self.send() cleanup
* extract Peer:new() from connect and accept
* fixup
* cleanup
It turns out that we drop connection if we fail to process a message
because of chain/store/internal error, eg we have a header already, so
we refuse it and drop the peer.
This pr doesn't forward this error to the peer error channel so the
connection will not be dropped.
* use a grin specific tmp dir instead of a system tmp dir
* rustfmt
* using create_new instead of create
* switch the txhashset sandbox from system tmp dir to grin specific tmp dir
* rustfmt
* adapter get_tmp_dir delegate to chain implementation
* simply the logic of protolc for get_tmpfile_pathname
* rustfmt
* fix the merge mistake
Most of the methods return nothing or bool which is used to decide if a
sender of a message should be banned or not. However underlying chain
implementation may fail so we need a way to reflect this fact in API.
Also it allows to reduce number of unwraps and makes the code more robust.
* cleanup legacy "3 dot" check (#2625)
* Allow to peers behind NAT to get up to preferred_max connections (#2543)
Allow to peers behind NAT to get up to preffered_max connections
If peer has only outbound connections it's mot likely behind NAT and we should not stop it from getting more outbound connections
* Reduce usage of unwrap in p2p crate (#2627)
Also change store crate a bit
* Simplify (and fix) output_pos cleanup during chain compaction (#2609)
* expose leaf pos iterator
use it for various things in txhashset when iterating over outputs
* fix
* cleanup
* rebuild output_pos index (and clear it out first) when compacting the chain
* fixup tests
* refactor to match on (output, proof) tuple
* add comments to compact() to explain what is going on.
* get rid of some boxing around the leaf_set iterator
* cleanup
* [docs] Add switch commitment documentation (#2526)
* remove references to no-longer existing switch commitment hash
(as switch commitments were removed in ca8447f3bd
and moved into the blinding factor of the Pedersen Commitment)
* some rewording (points vs curves) and fix of small formatting issues
* Add switch commitment documentation
* [docs] Documents in grin repo had translated in Korean. (#2604)
* Start to M/W intro translate in Korean
* translate in Korean
* add korean translation on intro
* table_of_content.md translate in Korean.
* table_of_content_KR.md finish translate in Korean, start to translate State_KR.md
* add state_KR.md & commit some translation in State_KR.md
* WIP stat_KR.md translation
* add build_KR.md && stratum_KR.md
* finish translate stratum_KR.md & table_of_content_KR.md
* rename intro.KR.md to intro_KR.md
* add intro_KR.md file path each language's intro.md
* add Korean translation file path to stratum.md & table_of_contents.md
* fix difference with grin/master
* Fix TxHashSet file filter for Windows. (#2641)
* Fix TxHashSet file filter for Windows.
* rustfmt
* Updating regexp
* Adding in test case
* Display the current download rate rather than the average when syncing the chain (#2633)
* When syncing the chain, calculate the displayed download speed using the current rate from the most recent iteration, rather than the average download speed from the entire syncing process.
* Replace the explicitly ignored variables in the pattern with an implicit ignore
* remove root = true from editorconfig (#2655)
* Add Medium post to intro (#2654)
Spoke to @yeastplume who agreed it makes sense to add the "Grin Transactions Explained, Step-by-Step" Medium post to intro.md
Open for suggestions on a better location.
* add a new configure item for log_max_files (#2601)
* add a new configure item for log_max_files
* rustfmt
* use a constant instead of multiple 32
* rustfmt
* Fix the build warning of deprecated trim_right_matches (#2662)
* [DOC] state.md, build.md and chain directory documents translate in Korean. (#2649)
* add md files for translation.
* start to translation fast-sync, code_structure. add file build_KR.md, states_KR.md
* add dandelion_KR.md && simulation_KR.md for Korean translation.
* add md files for translation.
* start to translation fast-sync, code_structure. add file build_KR.md, states_KR.md
* add dandelion_KR.md && simulation_KR.md for Korean translation.
* remove some useless md files for translation. this is rearrange set up translation order.
* add dot end of sentence & translate build.md in korean
* remove fast-sync_KR.md
* finish build_KR.md translation
* finish build_KR.md translation
* finish translation state_KR.md & add phrase in state.md to move other language md file
* translate blocks_and_headers.md && chain_sync.md in Korean
* add . in chain_sync.md , translation finished in doc/chain dir.
* fix some miss typos
* Api documentation fixes (#2646)
* Fix the API documentation for Chain Validate (v1/chain/validate). It was documented as a POST, but it is actually a GET request, which can be seen in its handler ChainValidationHandler
* Update the API V1 route list response to include the headers and merkleproof routes. Also clarify that for the chain/outputs route you must specify either byids or byheight to select outputs.
* refactor(ci): reorganize CI related code (#2658)
Break-down the CI related code into smaller more maintainable pieces.
* Specify grin or nanogrins in API docs where applicable (#2642)
* Set Content-Type in API client (#2680)
* Reduce number of unwraps in chain crate (#2679)
* fix: the restart of state sync doesn't work sometimes (#2687)
* let check_txhashset_needed return true on abnormal case (#2684)
* Reduce number of unwwaps in api crate (#2681)
* Reduce number of unwwaps in api crate
* Format use section
* Small QoL improvements for wallet developers (#2651)
* Small changes for wallet devs
* Move create_nonce into Keychain trait
* Replace match by map_err
* Add flag to Slate to skip fee check
* Fix secp dependency
* Remove check_fee flag in Slate
* Add Japanese edition of build.md (#2697)
* catch the panic to avoid peer thread quit early (#2686)
* catch the panic to avoid peer thread quit before taking the chance to ban
* move catch wrapper logic down into the util crate
* log the panic info
* keep txhashset.rs untouched
* remove a warning
* [DOC] dandelion.md, simulation.md ,fast-sync.md and pruning.md documents translate in Korean. (#2678)
* Show response code in API client error message (#2683)
It's hard to investigate what happens when an API client error is
printed out
* Add some better logging for get_outputs_by_id failure states (#2705)
* Switch commitment doc fixes (#2645)
Fix some typos and remove the use of parentheses in a
couple of places to make the reading flow a bit better.
* docs: update/add new README.md badges (#2708)
Replace existing badges with SVG counterparts and add a bunch of new ones.
* Update intro.md (#2702)
Add mention of censoring attack prevented by range proofs
* use sandbox folder for txhashset validation on state sync (#2685)
* use sandbox folder for txhashset validation on state sync
* rustfmt
* use temp directory as the sandbox instead actual db_root txhashset dir
* rustfmt
* move txhashset overwrite to the end of full validation
* fix travis-ci test
* rustfmt
* fix: hashset have 2 folders including txhashset and header
* rustfmt
*
(1)switch to rebuild_header_mmr instead of copy the sandbox header mmr
(2)lock txhashset when overwriting and opening and rebuild
* minor improve on sandbox_dir
* add Japanese edition of state.md (#2703)
* Attempt to fix broken TUI locale (#2713)
Can confirm that on the same machine 1.0.2 TUI looks great and is broken on
the current master. Bump of `cursive` version fixed it for me.
Fixes#2676
* clean the header folder in sandbox (#2716)
* forgot to clean the header folder in sandbox in #2685
* Reduce number of unwraps in servers crate (#2707)
It doesn't include stratum server which is sufficiently changed in 1.1
branch and adapters, which is big enough for a separate PR.
* rustfmt
* change version to beta
* reworked the dandelion rewrite (dandelion++)
* fallback to fluff/broadcast if we cannot stem the tx for any reason
* rework stem vs fluff logic during accepting tx
* cleanup docs
* add is_stem to logging
* cleanup
* rustfmt
* cleanup monitor and logging
* rework dandelion monitor to use simple cutoff for aggregation
* transition to next epoch *after* processing tx
so we fluff final outstanding txs
* fluff all txs in stempool if any are older than 30s
aggressively aggregate when we can
* fix rebase onto 1.1.0
* default config comments for Dandelion
* fix code to reflect our tests - fallback to txpool on stempool error
* log fluff and expire errors in dandelion monitor
* cleanup
* fix off by one
* cleanup
* cleanup
* various fixes
* one less clone
* cleanup
* expose leaf pos iterator
use it for various things in txhashset when iterating over outputs
* fix
* cleanup
* rebuild output_pos index (and clear it out first) when compacting the chain
* fixup tests
* refactor to match on (output, proof) tuple
* add comments to compact() to explain what is going on.
* get rid of some boxing around the leaf_set iterator
* cleanup
* dynamically resize lmdb
* rustfmt
* explicitly close db before resizing
* rustfmt
* test fix
* rustfmt
* pool tests
* chain fix
* merge
* move RwLock into Store, ensure resize gives a min threshold
* rustfmt
* move locks based on testing
* rustfmt
* wip
* big refactor, regretting doing this now
* PeerAddr everywhere
* cleanup
* fixup server tests
* peers api working for GET, POST is still WIP
* we can now ban/unban peers by ip only (port optional)
* fix: Fix race condition with dandelion_relay peer map and make more semantic
* Fix bug where we don't actually re-read the dandelion_relay variable after refreshing it
* Allow to have more than min_peers/2 outbound peers
Default value is 8, so we allow only 4 outbound peers, unfortunately if a node is behind NAT it can't get inbound peers and it stucks with 4 peers as maximum.
This PR allows a number of outbound peers to be between min_peers and max_peers/2. Node behind NAT will eventually have just min_peers number of peers.
* Extract into healthy_peers_mix function
* rework bucket txs
* introduce max_block_weight to global (based on chain_type)
not yet in use anywhere...
* now using global::max_block_weight() everywhere
* testing max_block_weight needs to be big enough to cover existing chain tests...
* add some test coverage for prepare mineable transactions at the block weight limit
introduce AsLimitedTx{max_weight} so we can build artifically small blocks (per mining config)
* cleanup
* default_mineable_max_weight is just the default max_block_weight
we do not need to account for coinbase reward here (tx vs block)
* 75 change outputs in a test is not valid now that we have a low block weight limit...
* outbound connection count
* rustfmt
* display outbound peer count
* rustfmt
* wip - allow connections to exceed max when trying to create new outcound connections
any in excess of total max will be cleaned up subsequently
* fix
* we care about connected outbound peer count when deciding to connect to new peers
* Refuse duplicate peer IPs
* Explicitly mentioning NAT in code comment
* Well of course that would mess with tests. Only trigger using IP test when on the wild internet (~DNS seeding).
* Break out of main peer loop on error
* Force client conn shutdown on error
* Fix borrow error
* Ban peers that fail handshake
* Fix add_peer for ban, remove useless disconnect
* shutdown the tcpstream when peer connect or accept fail
* remove the unnecessary 3 times retry on peer connecting
* connect/accept are actually handshakings instead of tcpstream connect/accept
* Fix secondary scaling bugs; rename is_testnet -> is_floonet (#2215)
* add global::is_mainnet()
* use it to change pre-genesis pow type
* rename is_testnet -> is_floonet
* Support multiple chain configurations (#2217)
* Support multiple chain configurations
Supports generating the proper configuration for each chain type
(mainnet, floonet, usernet). Will run them by default under
their respective root directory (~/.grin/main, ~/.grin/floo, etc).
Assigned default ports for mainnet, overriding them to keep Floonet
ports unchanged.
For now, starting on mainnet will abort.
* Fixed usernet command line help message. Fixes#2217
* Differing magic numbers for each chain type (#2208)
* stick to e=H(R|P|m) when use schnorr signature (#2200)
* stick to e=H(R|P|m) when use schnorr signature
* (1)add verify_slate_messages for wallet receive (2)log the message content
* remove debug log on verify_slate_messages
* verify the sender's message signature when receive_tx in wallet listen
* Revert "remove debug log on verify_slate_messages"
This reverts commit 65ea32a407.
* Revert "rustfmt"
This reverts commit c380ab9185.
* Revert "(1)add verify_slate_messages for wallet receive (2)log the message content"
This reverts commit 9584ca7a89.
* [re-floonet] Keychain Floonet BIP32 version/network option (#2235)
* add 'is_floonet' property to keychain
* fix hex encoding and tests
* Fix couple floonet loose ends (#2230)
* Fix couple floonet loose ends. Fixes#2216
* Doc fix for sig message
* Refuse unkown kernel features (#2244)
* Minor: magic number change for re-floonet
* Set pre genesis is_secondary to true (#2247)
* Minor: tx validation error display underlying
* New floonet genesis
* genesis rustfmt
* Use chain-specific config for wallet toml gen
* Fix default wallet_listener_url
* New more reasonable genesis block, bumped version
* genesis rustfmt
* Couple minor fixes to genesis generation script
* ban a peer with highest height but can't send us the proper headers
* reset self.stalling_ts if syncing goes well
* fix: switch to peers.ban_peer() instead of peer.set_banned()
* headers msg is now "streamed" off the tcp stream
* rustfmt
* cleanup
* move StreamingReader into ser.rs
extract read_exact out into util crate
* rustfmt
* do not treat txhashset.zip download as abusive behavior
* count times, not bytes so we exclude quiet increments
* use inc_quiet when tracking sent bytes via an attachment
* add comment
* fixup "quiet" counter entries
* rustfmt
* use FixedLength to define serialized size in bytes of various structs
replace usages of mem::size_of() with ::LEN so we correctly calculate serialized sizes in bytes
* rustfmt
* split horizon into two explicit values for cut through and txhashset request
* let node which has 2-7 days of history be able to handle forks larger than 2 days
* add test simulate_long_fork
* add pause/resume feature on p2p for tests
* refactor the state_sync
* ignore the test case simulate_long_fork for normal Travis-CI
* refactor function check_txhashset_needed to be shared with body_sync
* fix: state TxHashsetDone should allow header sync
* deserialize capabilities dropping unknown flag bits
* add tests for capabilities deserialization
* rustfmt
* FULL_NODE needs to be handled safely for legacy nodes
Fuzz test found that we don't read IPV6 addr (as part of p2p message)
properly. The code is supposed to read 8 dwords, but [0..8] is not a
slice of 8 ints, but a slice of one Range, so we always read just one
dword
* Update secp dependency to use crates.io version
* Add more details to various Cargo.toml
* Remove use of env variables that aren't reliably provided by cargo
* Initial expired peers removal
* Stop expired peers
* Simplify peer removal and remove only Defunct peers
* Make seed to check for expired peers every hour
* Get rid of unused vector of peers to remove
* Make peer deletion predicate closure immutable
* Replace logging backend to flexi-logger and add log rotation
* Changed flexi_logger to log4rs
* Disable logging level filtering in Root logger
* Support different logging levels for file and stdout
* Don't log messages from modules other than Grin-related
* Fix formatting
* Place backed up compressed log copies into log file directory
* Increase default log file size to 16 MiB
* Add comment to config file on log_max_size option
* Add peers used bandwidth calculation and display in TUI
* Fix formatting
* Change Mutex to RwLock from peer's used bandwidth statistics in Tracker
* Make used bandwidth column in TUI peers list sort by sum of bytes
* First pass at secondary proof of work difficulty adjustments
* Core and chain test fixes
* Next difficulty calc now needs a height. Scaling calculation fixes. Setting scaling on mined block.
* Change factor to u32 instead of u64.
* Cleanup structs used by next_difficulty
* Fix header size calc with u32 scaling
* fix: avoid a confusing log when fastsync start (#1720)
(cherry picked from commit 4f462cdfdc)
* small speed optimization for header sync (#1719)
(cherry picked from commit d3589d1bf5)
* get rid of RwLock around peer instances
* rustfmt
* rename stuff to live_info
* rustfmt
* serialize into PeerInfoDisplay
limit live_info write lock to peer_info.update()
rename to PeerLiveInfo
* simplify broadcast logic a bit more
connected peers are connected by definition
* return true/false on broadcast so we can send to more peers
* rustfmt
* fix p2p and server tests for PeerInfoDisplay
* commit
* Move Proof and Difficulty types to pow module. Difficulty scaling calculation.
* Diffculty scaling and size shift for 2nd PoW
* Backport e7eb26ee
* Test compilation fixes
* Scaling only need to go one way, as @tromp pointed out
Currently on startup we wait for 30 secs if we don't have at least 4 peers with more work. If a node was quickly restarted it is already fully synced so thereare no peers with more work at all. This pr assumes that if we've done some work already and we have enough peers but still 0 with more work we are fully synced and good to go.
* fix for unstable travis-ci test on servers module
* skip the PeerWithSelf connection request on sending side
* logs and comments change; and remove an unnecessary unwrap() in test
* improve: HeaderSync optimization (#1372)
* remove get_locator() optimization, which should be an independent pr for security review
* refactoring: move 'headers_streaming_body()' from Message to Protocol
* move 2 headers utils functions out of Protocol, and remove 'pub'
* support reading variable size of BlockHeader, from Cuckoo30 to Cuckoo36
* fix: use global::min_sizeshift() instead of hardcoded 30, because Cuckoo10 will be used for AutomatedTesting chain
* fix: should use global::proofsize() instead of hardcoded 42 when calculate serialized_size_of_header
* replace another 42 with global::proofsize()
