feat: add state size delta and trie depth#722
Open
weiihann wants to merge 6 commits into
Open
Conversation
weiihann
changed the title
Savid
reviewed
Jan 12, 2026
Author
|
Converted to draft since we have to implement the live tracer in geth first |
weiihann
force-pushed
the
feat/trie-depth
branch
from
93d3cbc to
2021e11
Compare
This was referenced Jan 14, 2026
weiihann
force-pushed
the
feat/trie-depth
branch
from
2021e11 to
e6c6f79
Compare
rename add parent root add bytes fix refactor: use map for depth feat(proto): add execution block state metrics feat(event-ingester): add execution block state metrics chore feat: add new module revert
…kHouse
The execution_state_size_delta table previously stored precomputed signed
deltas (account_delta, account_bytes_delta, ...). That representation lost
information: a block that updates one account looked the same as a block
that adds and removes accounts whose totals cancel.
This refactor splits each metric into a (writes, deletes) pair carrying
the gross churn:
- 20 stored Int64 columns covering 5 categories x {count, bytes} x
{writes, deletes}: accounts, account trie nodes, contract code,
storage slots, storage trie nodes.
- 10 MATERIALIZED Int64 columns derived as (writes - deletes) preserve
the original delta semantics for any query that wants the net change.
An "update" counts as both a write of the new value and a delete of the
prev value, so (writes - deletes) recovers the net delta for all three
cases (create / update / delete).
Contract code remains write-only in geth's state sizer (it is
content-addressed and shared across accounts, so deletion would require
reference counting). The contract_code_deletes / contract_code_delete_bytes
columns are present for schema symmetry and stay 0 until upstream geth
grows ref-counting; contract_code_delta therefore equals contract_code_writes.
Touches the geth tracer (writes/deletes nested JSON instead of delta),
the proto (renamed *_delta fields to *_writes and added *_deletes),
the Vector transforms (sentry-logs normaliser + kafka-to-clickhouse VRL),
and the ClickHouse migration. Verified end-to-end with the test-geth
1M-block mainnet import: 1,000,001 rows in each table, MATERIALIZED
delta identity holds for every row.
…igration
Both tables are populated from the same geth "State metrics" log line and
have no purpose without each other. Splitting them across two migration
files (003_execution_state_size_delta + 004_execution_mpt_depth) added
no value and made the ordering brittle when rebasing past unrelated
upstream migrations.
The combined 003_state_metrics.{up,down}.sql holds both CREATE/DROP
statements. The migrator's behaviour is unchanged — golang-migrate
splits on ';' and runs each statement individually, identical to running
two separate files.
Adds the hand-written and codegen pairs that let xatu-consumoor land
EXECUTION_STATE_SIZE_DELTA and EXECUTION_MPT_DEPTH events directly into
ClickHouse via the protobuf path.
Also fixes a chgo-rowgen omission: MATERIALIZED columns must not appear
in the insert column list. Previously the tool included them, which would
have caused INSERT to fail with "Cannot insert column ... — it is
MATERIALIZED". The fix filters default_kind IN ('MATERIALIZED', 'ALIAS')
from the system.columns query so the generated struct only carries
columns the consumer can actually write to. The fast_confirmation
.gen.go is regenerated as a side effect (only the map capacity hint
changed; no behavioural difference).
execution_mpt_depth.go also adapts the proto's map<uint32, uint64>
(proto3 has no uint8) to ClickHouse's Map(UInt8, UInt64) via toDepthMap.
Trie depths are bounded by [0, 64] per the geth state sizer; entries
with out-of-range keys are dropped defensively.
Verified end-to-end with the 1M-block test-geth import: 1,000,001 rows
in each table, the 10 MATERIALIZED delta identities hold for every row,
and the per-depth Map columns are populated correctly.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
In discussion with @samcm to utilize ethereum/go-ethereum@01b39c9 because there are issues with the
debug_stateSizemethod for state size collection.