The previous version of this change unconditionally returned the progress
map until the consensus client had driven the node at least once. That
broke ethclient.TestEthClient/StatusFunctions and any other backend that
runs without a consensus client (in-process tests, --dev mode without
catalyst, light/legacy backends), where reporting "syncing" forever is
clearly wrong.
Split the gate into two flags:
- clExpected: set in eth/catalyst.Register, the only entry point that
attaches the Engine API to a node. If a backend never calls Register,
it is not paired with a consensus client.
- clContacted: set on every Engine API call (forkchoiceUpdated and
newPayload), unchanged from before.
Replace ConsensusContacted on the Backend interface with ConsensusReady,
which folds the two flags into the question eth_syncing actually wants
answered: "is the synced claim meaningful right now?" Backends that
never expect a CL answer yes immediately, preserving legacy behavior.
Backends that do expect one answer yes only after the first FCU/NewPayload.
- eth/backend.go: clExpected, clContacted, MarkConsensusExpected,
MarkConsensusContacted, ConsensusReady on (*Ethereum)
- eth/catalyst/api.go: backend.MarkConsensusExpected() in Register
- eth/api_backend.go: ConsensusReady delegates to (*Ethereum)
- internal/ethapi/backend.go: rename interface method to ConsensusReady
- internal/ethapi/api.go: Syncing checks ConsensusReady
- internal/ethapi/{api_test,transaction_args_test}.go: rename the test
mock methods (default to true so existing tests are unaffected)
- internal/ethapi/syncing_test.go: rename the helper field; tests now
cover (a) CL-paired node before handshake -> truthy, (b) ready node
-> false, (c) active sync -> progress map regardless of gate
Refs #33687.
eth_syncing currently returns false as soon as the local downloader
believes the chain to be done. On a freshly started node this happens
before the consensus client has talked to it: the persisted head loads
into memory, no CL handshake has occurred, the downloader sees nothing
to do, Progress.Done() is true, eth_syncing reports synced.
That is wrong from an operator perspective. Load balancers (HAProxy,
NGINX), L2 supervisors and multi-node setups commonly gate routing on
eth_syncing. They start sending live traffic to a node that has not
actually learned about any new head yet, which surfaces as missing
state, stale reads, and unhealthy upstreams.
Maintainer-endorsed direction in the issue thread: "default geth to
'syncing' on startup and only switch to 'synced' once we learn about
a new block".
Implement that with a sticky atomic.Bool on *Ethereum, set the first
time the consensus layer drives the node via the Engine API
(ForkchoiceUpdated or NewPayload), and consulted from eth_syncing.
- eth/backend.go: add Ethereum.clContacted with
MarkConsensusContacted/ConsensusContacted helpers
- eth/catalyst/api.go: call MarkConsensusContacted at the same point
where lastForkchoiceUpdate / lastNewPayloadUpdate are stamped, so
the gate flips on every CL message regardless of the response
status (handshake recorded even when we reply STATUS_SYNCING)
- internal/ethapi/backend.go: add ConsensusContacted() to the Backend
interface and to the two test mocks (api_test.go testBackend,
transaction_args_test.go backendMock; both default to true so
existing tests keep their original semantics)
- eth/api_backend.go: implement ConsensusContacted on EthAPIBackend
- internal/ethapi/api.go: in EthereumAPI.Syncing, only short-circuit
to "false" when both progress.Done() AND ConsensusContacted() are
true; otherwise return the progress map as during an active sync
Adds dedicated tests in internal/ethapi/syncing_test.go covering:
- the new gate (Done but no CL contact -> truthy progress)
- normal post-handshake behavior (Done + CL contact -> false)
- active-sync behavior is unchanged regardless of the gate
Refs #33687.
Fixes https://github.com/ethereum/go-ethereum/issues/31732.
This logic was removed in the recent refactoring in the txindexer to
handle history cutoff (#31393). It was first introduced in this PR:
https://github.com/ethereum/go-ethereum/pull/28908.
I have tested it and it works as an alternative to #31745.
This PR packs 3 changes to the flow of fetching txs from the API:
- It caches the indexer tail after each run is over to avoid hitting the
db all the time as was done originally in #28908.
- Changes `backend.GetTransaction`. It doesn't return an error anymore
when tx indexer is in progress. It shifts the responsibility to the
caller to check the progress. The reason is that in most cases we anyway
check the txpool for the tx. If it was indeed a pending tx we can avoid
the indexer progress check.
---------
Co-authored-by: Gary Rong <garyrong0905@gmail.com>
This PR changes the chain view update mechanism of the log filter.
Previously the head updates were all wired through the indexer, even in
unindexed mode. This was both a bit weird and also unsafe as the
indexer's chain view was updates asynchronously with some delay, making
some log related tests flaky. Also, the reorg safety of the indexed
search was integrated with unindexed search in a weird way, relying on
`syncRange.ValidBlocks` in the unindexed case too, with a special
condition added to only consider the head of the valid range but not the
tail in the unindexed case.
In this PR the current chain view is directly accessible through the
filter backend and unindexed search is also chain view based, making it
inherently safe. The matcher sync mechanism is now only used for indexed
search as originally intended, removing a few ugly special conditions.
The PR is currently based on top of
https://github.com/ethereum/go-ethereum/pull/31642
Together they fix https://github.com/ethereum/go-ethereum/issues/31518
and replace https://github.com/ethereum/go-ethereum/pull/31542
---------
Co-authored-by: Gary Rong <garyrong0905@gmail.com>
The total difficulty is the sum of all block difficulties from genesis
to a certain block. This value was used in PoW for deciding which chain
is heavier, and thus which chain to select. Since PoS has a different
fork selection algorithm, all blocks since the merge have a difficulty
of 0, and all total difficulties are the same for the past 2 years.
Whilst the TDs are mostly useless nowadays, there was never really a
reason to mess around removing them since they are so tiny. This
reasoning changes when we go down the path of pruned chain history. In
order to reconstruct any TD, we **must** retrieve all the headers from
chain head to genesis and then iterate all the difficulties to compute
the TD.
In a world where we completely prune past chain segments (bodies,
receipts, headers), it is not possible to reconstruct the TD at all. In
a world where we still keep chain headers and prune only the rest,
reconstructing it possible as long as we process (or download) the chain
forward from genesis, but trying to snap sync the head first and
backfill later hits the same issue, the TD becomes impossible to
calculate until genesis is backfilled.
All in all, the TD is a messy out-of-state, out-of-consensus computed
field that is overall useless nowadays, but code relying on it forces
the client into certain modes of operation and prevents other modes or
other optimizations. This PR completely nukes out the TD from the node.
It doesn't compute it, it doesn't operate on it, it's as if it didn't
even exist.
Caveats:
- Whenever we have APIs that return TD (devp2p handshake, tracer, etc.)
we return a TD of 0.
- For era files, we recompute the TD during export time (fairly quick)
to retain the format content.
- It is not possible to "verify" the merge point (i.e. with TD gone, TTD
is useless). Since we're not verifying PoW any more, just blindly trust
it, not verifying but blindly trusting the many year old merge point
seems just the same trust model.
- Our tests still need to be able to generate pre and post merge blocks,
so they need a new way to split the merge without TTD. The PR introduces
a settable ttdBlock field on the consensus object which is used by tests
as the block where originally the TTD happened. This is not needed for
live nodes, we never want to generate old blocks.
- One merge transition consensus test was disabled. With a
non-operational TD, testing how the client reacts to TTD is useless, it
cannot react.
Questions:
- Should we also drop total terminal difficulty from the genesis json?
It's a number we cannot react on any more, so maybe it would be cleaner
to get rid of even more concepts.
---------
Co-authored-by: Gary Rong <garyrong0905@gmail.com>
This pull request refactors the EVM constructor by removing the
TxContext parameter.
The EVM object is frequently overused. Ideally, only a single EVM
instance should be created and reused throughout the entire state
transition of a block, with the transaction context switched as needed
by calling evm.SetTxContext.
Unfortunately, in some parts of the code, the EVM object is repeatedly
created, resulting in unnecessary complexity. This pull request is the
first step towards gradually improving and simplifying this setup.
---------
Co-authored-by: Martin Holst Swende <martin@swende.se>
* miner: untangle miner
* miner: use common.hash instead of *types.header
* cmd/geth: deprecate --mine
* eth: get rid of most miner api
* console: get rid of coinbase in welcome message
* miner/stress: get rid of the miner stress test
* eth: get rid of miner.setEtherbase
* ethstats: remove miner and hashrate flags
* ethstats: remove miner and hashrate flags
* cmd: rename pendingBlockProducer to miner.pending.feeRecipient flag
* miner: use pendingFeeRecipient instead of etherbase
* miner: add mutex to protect the pending block
* miner: add mutex to protect the pending block
* eth: get rid of etherbase mentions
* miner: no need to lock the coinbase
* eth, miner: fix linter
---------
Co-authored-by: Martin Holst Swende <martin@swende.se>
Co-authored-by: Péter Szilágyi <peterke@gmail.com>
This change simplifies the logic for indexing transactions and enhances the UX when transaction is not found by returning more information to users.
Transaction indexing is now considered as a part of the initial sync, and `eth.syncing` will thus be `true` if transaction indexing is not yet finished. API consumers can use the syncing status to determine if the node is ready to serve users.
EIP-4844 adds a new transaction type for blobs. Users can submit such transactions via `eth_sendRawTransaction`. In this PR we refrain from adding support to `eth_sendTransaction` and in fact it will fail if the user passes in a blob hash.
However since the chain can handle such transactions it makes sense to allow simulating them. E.g. an L2 operator should be able to simulate submitting a rollup blob and updating the L2 state. Most methods that take in a transaction object should recognize blobs. The change boils down to adding `blobVersionedHashes` and `maxFeePerBlobGas` to `TransactionArgs`. In summary:
- `eth_sendTransaction`: will fail for blob txes
- `eth_signTransaction`: will fail for blob txes
The methods that sign txes does not, as of this PR, add support the for new EIP-4844 transaction types. Resuming the summary:
- `eth_sendRawTransaction`: can send blob txes
- `eth_fillTransaction`: will fill in a blob tx. Note: here we simply fill in normal transaction fields + possibly `maxFeePerBlobGas` when blobs are present. One can imagine a more elaborate set-up where users can submit blobs themselves and we fill in proofs and commitments and such. Left for future PRs if desired.
- `eth_call`: can simulate blob messages
- `eth_estimateGas`: blobs have no effect here. They have a separate unit of gas which is not tunable in the transaction.
* all: move main transaction pool into a subpool
* go.mod: remove superfluous updates
* core/txpool: review fixes, handle txs rejected by all subpools
* core/txpool: typos
Adds an optional config parameter to eth_call which allows users to override block context fields (same functionality that was added to traceCall in #24871)
---------
Co-authored-by: Martin Holst Swende <martin@swende.se>
Here, the core.Message interface turns into a plain struct and
types.Message gets removed.
This is a breaking change to packages core and core/types. While we do
not promise API stability for package core, we do for core/types. An
exception can be made for types.Message, since it doesn't have any
purpose apart from invoking the state transition in package core.
types.Message was also marked deprecated by the same commit it
got added in, 4dca5d4db7 (November 2016).
The core.Message interface was added in December 2014, in commit
db494170dc, for the purpose of 'testing' state transitions. It's the
same change that made transaction struct fields private. Before that,
the state transition used *types.Transaction directly.
Over time, multiple implementations of the interface accrued across
different packages, since constructing a Message is required whenever
one wants to invoke the state transition. These implementations all
looked very similar, a struct with private fields exposing the fields
as accessor methods.
By changing Message into a struct with public fields we can remove all
these useless interface implementations. It will also hopefully
simplify future changes to the type with less updates to apply across
all of go-ethereum when a field is added to Message.
---------
Co-authored-by: Felix Lange <fjl@twurst.com>
Logs stored on disk have minimal information. Contextual information such as block
number, index of log in block, index of transaction in block are filled in upon request.
We can fill in all these fields only having the block header and list of receipts.
But determining the transaction hash of a log requires the block body.
The goal of this PR is postponing this retrieval until we are sure we the transaction hash.
It happens often that the header bloom filter signals there might be matches in a block,
but after actually checking them reveals the logs do not match. We want to avoid fetching
the body in this case.
Note that this changes the semantics of Backend.GetLogs. Downstream callers of
GetLogs now assume log context fields have not been derived, and need to call
DeriveFields on the logs if necessary.
