* remove max pivot block resets during fast sync
* increase max retry number and fix test
* change logs in the handleFailure method
* change logs related to suspicious number of retries
Signed-off-by: Karim TAAM <karim.t2am@gmail.com>
* Fix strategy for getting next nonce for sender in `PendingTransactions`.
Added test.
Signed-off-by: Abdelhamid Bakhta <abdelhamid.bakhta@consensys.net>
* Introduced `TransactionsForSenderInfo` structure to have a less naive implementation, more robust against load testing.
A queue of gaps between used nonces is maintained.
`getNextNonceForSender` has been updated and have the following logic:
- if no infos corresponding to the nonce returns an empty optional
- if gap list is not empty returns the lowest nonce in the gap
- else returns the max nonce + 1
Signed-off-by: Abdelhamid Bakhta <abdelhamid.bakhta@consensys.net>
* spotless
Signed-off-by: Abdelhamid Bakhta <abdelhamid.bakhta@consensys.net>
* spotless
Signed-off-by: Abdelhamid Bakhta <abdelhamid.bakhta@consensys.net>
* added test
Signed-off-by: Abdelhamid Bakhta <abdelhamid.bakhta@consensys.net>
* spotless
Signed-off-by: Abdelhamid Bakhta <abdelhamid.bakhta@consensys.net>
Removes as many Gradle 7.0 compatibility issues as possible
* `baseName` -> `archiveBaseName`
* `extension` -> `archiveExtension`
* `destinationDir` -> `destinationDirectory`
* `runtime` -> `runtimeOnly`
* Change some log4j-api and log4j-core dependencies
* Remove an unneeded and outdated plugin (`net.ltgt.apt`)
* tweak the plugin-api change detector's property annotations.
Warnings still exist with one external plugin used for license file
checking that we do not control the source code for.
Signed-off-by: Danno Ferrin <danno.ferrin@gmail.com>
This was turned off a number of months ago because of long build times.
However it appears either because of code structure changes or fixed
errorprone that this finishes in a reasonable time now.
Code that violated this check is also fixed.
Signed-off-by: Danno Ferrin <danno.ferrin@gmail.com>
Update dependencies to most current version
- except picocli which is a major version update
Alphabetize dependencies
Signed-off-by: Danno Ferrin <danno.ferrin@gmail.com>
It was identified that when both KV_ROCKSDB metrics and Privacy were enabled,
that Besu failed to start due to a naming collision in Metrics Collators - this was ultimately due to Besu having 2 keyvalue stores - one for public state, and another for private state - and _both_ using the same metrics.
To overcome this issue, the metrics used in the private kv store are prefixed with the word "private".
Signed-off-by: Trent Mohay <trent.mohay@consensys.net>
This includes code to address reading config for classic network and
kotti test network.
ClassicForkPeerValidator checks peers at classicForkBlock (same as Dao
Fork Block) and connects with peers that did not do the Dao Fork and
continued on the classic network.
Signed-off-by: edwardmack <ed@edwardmack.com>
Upgrade dependencies except rocksdb (needs burn in testing),
picocli (reorders options), gradle (causes build server breakage), and
web3j (test failures).
* Awaitality removed a Duration object and instead uses java.time
* jackson stopped throwing a checked exception for one API
* spotless now enforces gradle formatting checks (yea!)
Signed-off-by: Danno Ferrin <danno.ferrin@gmail.com>
* Add totalDiffculty to BlockPropagated events.
The chain head block can be contentious with many fork blocks (ommers) propagating on the network. We should add a totalDifficulty to make it easier to see which block is most likely the current head.
- added `BlockPropagated` interface in `plugin-api`.
- updated `BesuEvents.onBlockPropagated` method to take a `BlockPropagated` instead of a `BlockHeader`.
- created `BlockPropagatedSubscriber` in `BlockBroadcaster`.
- changed type of `BlockBroadcaster.blockPropagatedSubscribers` from `Consumer<Block>` to `BlockPropagatedSubscriber`.
- updated unit tests accordingly to all changes.
- updated known hash in `build.gradle` file of `plugin-api`: new value is `4SAeaZIJMsDvUK5Wp2RzU8TlHacslALnM/4yvVhsMtY=`
Signed-off-by: Abdelhamid Bakhta <abdelhamid.bakhta@consensys.net>
Fixes race condition between attaching the NodeAddedListener and some state from post marked block being persisted. We now prepare and cleanup the listener only once each, on start and stop respectively.
Other changes:
Start the Pruner before FullSyncDownloader. Luckily the downloader took enough time to start downloading that this wasn't an issue but it's safer this way.
Signed-off-by: Ratan Rai Sur <ratan.r.sur@gmail.com>
Add a new CLI flag `--required-block` that takes a block number and a
block hash. Before using a peer for syncing we validate that the block
exists with the spcified hash at the peer.
For example `--required-block=6485846=0x43f0cd1e5b1f9c4d5cda26c240b59ee4f1b510d0a185aa8fd476d091b0097a80`
deals with the current Istanbul Ropsten chainsplit.
Signed-off-by: Danno Ferrin <danno.ferrin@gmail.com>
We should not send a sync status for every forking block state update.
Yes, we send status updates for detected forks as well as new canonical
heads.
Instead we should send a synching message for status changes as well as
when we reorg the chain.
Signed-off-by: Danno Ferrin <danno.ferrin@gmail.com>
* adding in spdx-license-identifier & updated check for the same; removing license check from spotless
Signed-off-by: Joshua Fernandes <joshua.fernandes@consensys.net>
* Change CheckSpdxHeader to a task.
Signed-off-by: Adrian Sutton <adrian.sutton@consensys.net>
Expose EthPeer validation state so that the Synchronizer can choose peers based on whether or not they have been fully validated. This allows us to use only fully validated peers when choosing a pivot block.
Signed-off-by: Meredith Baxter <meredith.baxter@consensys.net>
Signed-off-by: Danno Ferrin <danno.ferrin@gmail.com>
Interally we use a `BigInteger` for ChainID, but currently use an `int` for
NetworkID. Because the default for NetworkID is the same value as ChainID there
are some chains where this will be very problematic, and there is at least
one other long-living chain outside the 32 bit int value range.
This is a large commit because the type is baked fairly deep into some of
the other APIs and very much in the test code testing the type.
Signed-off-by: Adrian Sutton <adrian.sutton@consensys.net>
As part of the plugin events service we need to support adding and removing
listeners for events. TransactionPool only supports adding pending and dropped
transactions. To support removing we need to pass around the subscription id.
To make it consistent some APIs were renamed.
Signed-off-by: Adrian Sutton <adrian.sutton@consensys.net>
* Mark Sweep Pruner
* add `unload` method on Node interface, which is a noop everywhere but on `StoredNode`
Signed-off-by: Adrian Sutton <adrian.sutton@consensys.net>
This reverts commit 814b36e4
The needed chantes to get rid of Instant.now (which is also needed to get rid
of the wall clock dependency) are too deep and intrusive into IBFT to try and
speed patch them in that some APIs require re-work, so in the interst of test
stability this gets sheleved until it is all ready.
Signed-off-by: Adrian Sutton <adrian.sutton@consensys.net>
To allow us to reset the timestamp in the blockchain for Retesteth support
we need to pass a Clock to affected APIs and use that instead of the static method
System.currentTimeMillis(). The most consistent way to do this that will ensure
that the API does not sneak back in is to ban the method via ErrorProne.
TestClock.fixed() was altered to return the "now" time of the first time the fixed clock was requested, needed for many header validation tasks validating headers are not from the future.
Signed-off-by: Adrian Sutton <adrian.sutton@consensys.net>
Karim T
Danno Ferrin
Abdelhamid Bakhta
Antoine Toulme
Steve Mieskoski
Trent Mohay
Edward Mack
mbaxter
Ratan Rai Sur
Jason Frame
Adrian Sutton
Joshua Fernandes
CJ Hare
mark-terry
Danno Ferrin
mbaxter
Karim T