The per-call SERVER span ended inside `handleCall()`, so the JSON-RPC
response write happened after the span closed. For large responses like
`engine_getBlobsV*`, that write time was missing from traces.
- Extend the SERVER span past `writeJSON`.
- For batches, add a top-level `jsonrpc.batch` SERVER span (with `rpc.batch.size`) covering the whole batch including `callBuffer.write`.
- Add `rpc.writeJSON` span around the non-batch response write.
- Add `rpc.writeJSONBatch` span around the batch response write.
- Add `rpc.httpWrite` span around the actual HTTP write, separating JSON encoding from network write.
- Add additional telemetry helpers.
---------
Co-authored-by: Felix Lange <fjl@twurst.com>
This PR is a prerequisite for landing snap v2, the BAL-healing snap sync
algorithm.
It duplicates much of the snap v1 skeleton, which is expected to be
deprecated once v2 is enabled. The code duplication is acceptable as a
short-term tradeoff, simplifying development and reducing integration
complexity.
This PR implements flat-file storage for finalized block access lists,
specifically:
* The freezer is extended with the notion of tail groups, allowing
different groups within a single freezer instance to maintain
independent tails, while all tables within the same group remain
tail-aligned.
* The freezer can now dynamically attach new tables to an existing
freezer instance, with both the table head and tail initialized to the
freezer's common head.
* A new freezer table, **bals**, has been added to the chain freezer
with its own dedicated tail group, preserving the flexibility to deploy
a tail-pruning policy different from the main chain data group.
Additionally, the BALs in the key-value store will be migrated to the
freezer instance once they are finalized or there are at least 90K block
confirmations on top acting as a "soft finalization". This freezing
policy is same with all chain segment data.
Checks the Ledger Ethereum app version before sending typed transactions
that require newer app support.
EIP-2930/EIP-1559 transactions now require Ledger app v1.9.0 or newer,
and EIP-7702 transactions require v1.17.0 or newer. Older apps now
return the same kind of local update error already used for earlier
Ledger feature gates instead of sending an unsupported transaction to
the device.
---------
Co-authored-by: Guillaume Ballet <3272758+gballet@users.noreply.github.com>
Because the UBT doesn't differentiate slots from accounts, the content
of the tree can not be exported as a `GenesisAlloc`, which means that
`evm t8n` can not intergrate it. We have tried integrating the new
format into execution-specs, but this is very hard to maintain because
the team doesn't see it as a priority and their own repository is seeing
a lot of churn. This PR adds the ability to capture the structure of
what is being inserted in the tree, so that the information isn't lost
and it can be dumped in the t8n context.
---------
Co-authored-by: felipe <fselmo2@gmail.com>
The server already rejects empty batches with -32600. On the client
side, calling BatchCallContext with a zero-length slice on inproc/WS/IPC
transports registers no request IDs but the server still replies with an
error message whose id is null. The dispatch loop has no requestOp to
match it to, so op.resp is never written and op.wait blocks until ctx
deadline.
Short-circuit on len(b) == 0 with the same invalidRequestError the
server uses, so all transports return immediately with -32600.
Add a disableGzip parameter to NewHTTPHandlerStack and httpConfig.
initAuth sets it true so compression is disabled in the engine api.
Public HTTP RPC behavior is unchanged.
This is another one of my slop-PRs, aimed at reducing the amount of
future slop PRs by doing it all in one go.
All of the deprecated cli flags have been in that state for over a year.
It's time to remove them, especially since they are ineffective.
Note that I kept the code to report and manage deprecated cli flags, as
I assume we will be deprecating more flags in the future.
### Summary
Closes#34621.
`github.com/pion/dtls/v2` is affected by
[CVE-2026-26014](https://nvd.nist.gov/vuln/detail/CVE-2026-26014); the
fix lives in `github.com/pion/dtls/v3`. In this tree, dtls/v2 is pulled
in indirectly via `github.com/pion/stun/v2 v2.0.0` (declared at
`go.mod:53`), which is the only direct consumer — `p2p/nat/stun.go` is
the sole call site.
`github.com/pion/stun/v3` already uses dtls/v3, so bumping `stun`
upgrades the vulnerable dependency without touching `pion/dtls`
directly.
### API check
The v3 surface used by `p2p/nat/stun.go` is byte-identical in shape to
v2:
| Symbol | v2 | v3 |
|---|---|---|
| `Dial` | `func Dial(network, address string) (*Client, error)` | same
|
| `Build` | `func Build(setters ...Setter) (*Message, error)` | same |
| `TransactionID` | `var TransactionID Setter` | same |
| `BindingRequest` | `var BindingRequest = NewType(MethodBinding,
ClassRequest)` | same |
| `Event` | `type Event struct` | same |
| `XORMappedAddress` | `type XORMappedAddress struct { …
GetFrom(*Message) error }` | same |
| `DefaultPort` | `const DefaultPort = 3478` | same |
So the code change is just the import rename plus an alias rename to
keep the local label honest (`stunV2` → `stunV3`).
### Change
`go.mod` / `go.sum`:
- Replace direct `github.com/pion/stun/v2 v2.0.0` with
`github.com/pion/stun/v3 v3.0.1`.
- `go mod tidy` drops every `pion/dtls/v2` and `pion/stun/v2` entry from
`go.sum` and pulls `pion/dtls/v3 v3.0.7`, `pion/stun/v3 v3.0.1`,
`pion/transport/v3 v3.0.8` as the new indirect set.
`p2p/nat/stun.go`:
- Update the import path and rename the alias from `stunV2` to `stunV3`.
### Verification
- `go build ./p2p/nat/` clean.
- `go test ./p2p/nat/ -count=1` passes (26s).
- `grep 'pion/dtls/v2\|pion/stun/v2' go.sum` returns zero matches.
### Notes
- `pion/dtls` is not imported directly anywhere in the tree, so no other
code needs touching.
- `pion/transport/v3` was already in the dependency graph (the `stun/v3`
upgrade just bumps the patch from v3.0.1 → v3.0.8); the v2 transport
drops out cleanly.
fixes#32672
This is kind of a band aid solution since it fixes the issue by
bypassing the snap sync expectations of an empty db and attempting to
import the new payload if we're at block 1. The next FCU will set the
status to synced.
Will continue looking to better understand how the above issue arises
and find a more thorough solution.
---------
Co-authored-by: Marius van der Wijden <m.vanderwijden@live.de>
Fixes#35033
## Problem
The GraphQL HTTP handler decoded request bodies directly before
executing the query. Unlike the JSON-RPC HTTP path, `/graphql` did not
have an explicit request body limit before JSON decoding.
A single `Decode` also stops after the first JSON value, so the handler
now requires EOF after the GraphQL request object to ensure oversized
trailing request data is not ignored.
## Changes
- Limit GraphQL request bodies to 5 MiB, matching the existing JSON-RPC
default body limit.
- Return `413 Request Entity Too Large` when the limit is exceeded.
- Require EOF after the request JSON object.
- Add regression coverage for oversized query bodies and oversized
trailing request data.
- Fix an existing GraphQL test fixture that had an unintended trailing
quote after the JSON object.
## Validation
- `gofmt -w graphql/service.go graphql/graphql_test.go`
- `go run golang.org/x/tools/cmd/goimports@latest -w graphql/service.go
graphql/graphql_test.go`
- `go test ./graphql -run TestGraphQLHTTPBodyLimit -count=1`
- `go test ./graphql -count=1`
There is currently no way for JSON-RPC clients to discover which
historical data a node can serve without probing with trial-and-error
calls and interpreting opaque error messages (`pruned history
unavailable`).
This makes it hard to build robust tooling on top of nodes that prune
their history, for example nodes started with `--history.chain
postmerge`
or with reduced `TransactionHistory`, `LogHistory`, or `StateHistory`
windows.
This PR implements `eth_capabilities` as defined in
ethereum/execution-apis#755. The method takes no parameters and returns
the current head plus six per-resource capability records:
- `state`
- `tx`
- `logs`
- `receipts`
- `blocks`
- `stateproofs`
Closes#33828
Archive nodes store the full history of transactions in the index. This PR
fixes a bug for users who provided the NoPruning field in a YAML config file.
Now geth correctly stores full transaction history if archive is configured via
YAML.
---------
Co-authored-by: Sina Mahmoodi <itz.s1na@gmail.com>
Each commit on a PR kicks off a CI run. Those CI jobs run to the finish
regardless, even when new commits have been pushed which make them stale
and useless. This change attempts to cancel any previously running job
for the same PR.
Passing `--state.size-tracking=false` currently cannot disable state
size tracking when it was enabled by the config file because the CLI
path only turns the config value on.
---------
Co-authored-by: Jared Wasinger <j-wasinger@hotmail.com>
The first NODES response sets total = min(int(response.RespCount),
totalNodesResponseLimit), With RespCount=0, total=0 but receive become
1; receive == count is never satisfied.
Return empty raw bytes when the GraphQL `Block.raw` resolver cannot load
the block body. This matches the nil handling used by the other
block-body-backed resolvers and avoids exposing RLP empty-list bytes as
raw block data.
The changes here enable us to fill tests with Amsterdam using geth EVM
bin.
This will be useful for block builder tests using `testing_buildBlockV1`
endpoint and for filling benchmarking compute and stateful tests as
Python is too slow for benchmark tests.
Tested in
[ethereum/execution-specs](https://github.com/ethereum/execution-specs)
with:
```
uv run fill --clean --fork=Amsterdam tests/amsterdam/eip7928_block_level_access_lists/test_block_access_lists.py --evm-bin=$GETH_EVM_PATH
```
- Adds tracing to the `GetBlobsV1/V2/V3`
- Adds `blobs.requested` and `blobs.filled` attributes to
`GetBlobsV1/V2/V3` spans.
- Adds tracing to `BlobTxPool().GetBlobs()`
This PR:
- Adds `engine_newPayloadWithWitnessV5`. The codebase already supports
the previous `VX`, so only `V5` was missing.
- Make the consensus witness format use the field [ordering defined in
the
spec](8d7e68f4b7/src/ethereum/forks/amsterdam/stateless_host_exec_witness.py (L175-L176))
to make it canonical.
cc @gballet
---------
Co-authored-by: Guillaume Ballet <3272758+gballet@users.noreply.github.com>
Fixes a regression where nil results from getBlobs were encoded as an empty array instead of null.
---------
Co-authored-by: Felix Lange <fjl@twurst.com>
