Separate BAL from block body by mirgee · Pull Request #9629 · besu-eth/besu

mirgee · 2026-01-13T11:10:53Z

With these changes, BALs are removed from the block body, stored and handled separately for both PoS and PoA networks.

The follow up PRs will:

add a CLI option to allow configurable retention policy for BALs (we must make sure to never prune BALs younger than WSP)
remove method overloads w/o BALs to make BALs an explicit argument (optional)

Testing was done by running pandoras-box against QBFT network with and without BALs and by running a local devnet across Amsterdam fork boundary using Ethereum Kurtosis package with tx_fuzz and spamoor.

Reference tests bal@v2.0.0 are passing, bal@v3.0.1 are not (due to incorrect state roots - maybe a problem with BPO config?).

Closes: #9589

Signed-off-by: Miroslav Kovar <miroslavkovar@protonmail.com>

…rgument Signed-off-by: Miroslav Kovar <miroslavkovar@protonmail.com>

Signed-off-by: Miroslav Kovar <miroslavkovar@protonmail.com>

… QBFT is active Signed-off-by: Miroslav Kovar <miroslavkovar@protonmail.com>

Signed-off-by: Miroslav Kovar <miroslavkovar@protonmail.com>

Signed-off-by: Karim Taam <karim.t2am@gmail.com>

.../src/main/java/org/hyperledger/besu/consensus/common/bft/BaseBftProtocolScheduleBuilder.java

.../main/java/org/hyperledger/besu/ethereum/referencetests/BlockchainReferenceTestCaseSpec.java

...reum/core/src/main/java/org/hyperledger/besu/ethereum/transaction/BlockSimulationResult.java

...qbft-core/src/main/java/org/hyperledger/besu/consensus/qbft/core/statemachine/QbftRound.java

consensus/ibft/src/main/java/org/hyperledger/besu/consensus/ibft/messagewrappers/Proposal.java

Signed-off-by: Karim Taam <karim.t2am@gmail.com>

matkt · 2026-01-19T14:33:35Z

Started 2 nodes on mainnet to check if there is performance or stability issue with this PR
and also 1 fullsync on hoodi

Signed-off-by: Miroslav Kovar <miroslavkovar@protonmail.com>

matthew1001 · 2026-01-21T16:58:38Z

...e/src/main/java/org/hyperledger/besu/consensus/qbft/core/network/QbftMessageTransmitter.java

  public void multicastProposal(
      final ConsensusRoundIdentifier roundIdentifier,
      final QbftBlock block,
+      final Optional<BlockAccessList> blockAccessList,


I could do with a little more context/background to the PR that would help me understand the changes here.

So my understanding of BALs is they are a new field in the block header? But your PR separates them out from the block.

In your PR you've said stored and handled separately for both PoS and PoA networks. I think I understand storing them separately on disk - to have different retention policies etc. But I'm not sure why the code needs to pass them around separately to the block? We currently pass around QbftBlock - is there a reason this can't just contain an Optional BAL field rather than changing all of the internal function signatures?

So my understanding of BALs is they are a new field in the block header? But your PR separates them out from the block.

There is a new BAL-related field in the block header called balHash. BALs used to be part of the block body (and hence stored together). This PR mainly separates BALs from the block body.

Passing BALs in the block was actually an intermediate stage when working on this PR, but then I invested into passing it around as a separate argument, because, conceptually, a block consists of only body + header, not body + header + BAL. BAL is a data structure separate from the block.

I don't fully understand the intended purpose of QbftBlock, but currently its only implementation simply wraps a Besu block (body + header) and wrapped Besu header (again). If that is the case, not coupling QbftBlock with BALs makes it potentially easier to remove (and thus simplify the code).

Signed-off-by: Miroslav Kovar <miroslavkovar@protonmail.com> # Conflicts: # consensus/qbft-core/src/test/java/org/hyperledger/besu/consensus/qbft/core/messagewrappers/ProposalTest.java # ethereum/core/src/main/java/org/hyperledger/besu/ethereum/MainnetBlockValidator.java

mirgee · 2026-01-22T08:15:59Z

Tested last version via kurtosis again, seems running fine. However currently there does seem to be a regression in QBFT.

EDIT: Fixed in fe37907

Signed-off-by: Miroslav Kovar <miroslavkovar@protonmail.com>

matkt

LGTM,

Tested on mainnet, hoodi and also tried to produce block on hoodi with a validator

Signed-off-by: Miroslav Kovar <miroslavkovar@protonmail.com> Co-authored-by: Karim Taam <karim.t2am@gmail.com>

…nodes PR besu-eth#9629 added blockAccessList to QBFT/IBFT consensus messages but broke backward compatibility with older nodes that do not include this field. PR besu-eth#9977 partially fixed this for QBFT ProposalPayload (where blockAccessList is the last field, so isEndOfCurrentList() suffices), but missed QBFT RoundChange where blockAccessList sits *before* the Prepares list. In that position isEndOfCurrentList() returns false (because Prepares is still there), causing the decoder to consume the Prepares list as a BlockAccessList and then fail with: RLPException: Cannot enter a lists, input is fully consumed Fix: use the enterList() item count to detect old (3-item) vs new (4-item) format in QBFT RoundChange.decode(). Also add the missing isEndOfCurrentList() guard to IBFT RoundChange and IBFT Proposal (where blockAccessList is the last field). Signed-off-by: qinglin <qinglin@example.com> 🤖 Generated with [Qoder][https://qoder.com]

mirgee added 5 commits January 13, 2026 12:17

Add BAL prefix and adjust CRUD API

ed8d179

Signed-off-by: Miroslav Kovar <miroslavkovar@protonmail.com>

Remove BALs from block body, adjust interfaces and pass as optional a…

c3bd64d

…rgument Signed-off-by: Miroslav Kovar <miroslavkovar@protonmail.com>

Handle BALs alongside block in BFT protocols

f49e6d6

Signed-off-by: Miroslav Kovar <miroslavkovar@protonmail.com>

Align block formatting (withdrawals presence) with fork activation if…

bbe0e34

… QBFT is active Signed-off-by: Miroslav Kovar <miroslavkovar@protonmail.com>

Include block access list in debug_getBadBlocks result

b8259ff

Signed-off-by: Miroslav Kovar <miroslavkovar@protonmail.com>

mirgee force-pushed the separate-bals-from-block-body branch from 3d91f05 to 9fe66cf Compare January 13, 2026 11:17

Improvements and simplifications

4872e93

Signed-off-by: Miroslav Kovar <miroslavkovar@protonmail.com>

mirgee force-pushed the separate-bals-from-block-body branch from 9fe66cf to 4872e93 Compare January 13, 2026 18:10

mirgee mentioned this pull request Jan 14, 2026

Configurable BAL retention policy #9637

Merged

mirgee force-pushed the separate-bals-from-block-body branch from fc60c5e to cec133c Compare January 15, 2026 09:27

Fix build and tests

0e2dd4a

Signed-off-by: Miroslav Kovar <miroslavkovar@protonmail.com>

mirgee force-pushed the separate-bals-from-block-body branch from cec133c to 0e2dd4a Compare January 15, 2026 15:21

mirgee added 2 commits January 15, 2026 17:31

Use BAL in the reference tests

11d07f9

Signed-off-by: Miroslav Kovar <miroslavkovar@protonmail.com>

Allow processing blocks without BAL but with balHash

14288ee

Signed-off-by: Miroslav Kovar <miroslavkovar@protonmail.com>

mirgee mentioned this pull request Jan 16, 2026

Add eth_getBlockAccessListBy* methods #9643

Merged

fix invalid unit tests when BAL is missing

efbc043

Signed-off-by: Karim Taam <karim.t2am@gmail.com>

matkt reviewed Jan 19, 2026

View reviewed changes

consensus/ibft/src/main/java/org/hyperledger/besu/consensus/ibft/messagewrappers/Proposal.java Outdated Show resolved Hide resolved

add qbft tests and cleaning

9116582

Signed-off-by: Karim Taam <karim.t2am@gmail.com>

Address code review comments

a701f40

Signed-off-by: Miroslav Kovar <miroslavkovar@protonmail.com>

matthew1001 approved these changes Jan 21, 2026

View reviewed changes

macfarla moved this to Open PRs in 26.1.0 Release Jan 22, 2026

macfarla added this to 26.1.0 Release Jan 22, 2026

Use noop request processor coordinator for PoA

fe37907

Signed-off-by: Miroslav Kovar <miroslavkovar@protonmail.com>

matkt approved these changes Jan 22, 2026

View reviewed changes

Merge branch 'main' into separate-bals-from-block-body

2f7cbdc

matkt enabled auto-merge (squash) January 22, 2026 10:28

matkt merged commit a07c7bf into besu-eth:main Jan 22, 2026
46 of 61 checks passed

github-project-automation bot moved this from Open PRs to Done in 26.1.0 Release Jan 22, 2026

macfarla pushed a commit to CPerezz/besu that referenced this pull request Feb 6, 2026

Separate BAL from block body (besu-eth#9629)

59b4b1d

Signed-off-by: Miroslav Kovar <miroslavkovar@protonmail.com> Co-authored-by: Karim Taam <karim.t2am@gmail.com>

matthew1001 mentioned this pull request Feb 17, 2026

QBFT FullSync fails after upgrade: new/wiped StatefulSet pod cannot sync blocks produced by 25.12.0 with Shanghai enabled #9786

Closed

mirgee mentioned this pull request Feb 17, 2026

Reintroduce NotApplicableWithdrawals withdrawals validator #9830

Merged

matkt mentioned this pull request Mar 12, 2026

Propose Miroslav Kovář as a maintainer #10033

Merged

daniellehrner mentioned this pull request Apr 1, 2026

Update block processor method used in Besu Consensys/linea-monorepo#2706

Merged

4 tasks

ghostant-1017 mentioned this pull request Apr 9, 2026

Fix QBFT/IBFT RLPException when decoding RoundChange from pre-26.1.0 nodes #10204

Open

5 tasks

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Separate BAL from block body#9629

Separate BAL from block body#9629
matkt merged 15 commits intobesu-eth:mainfrom
mirgee:separate-bals-from-block-body

mirgee commented Jan 13, 2026 •

edited

Loading

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

matkt commented Jan 19, 2026

Uh oh!

matthew1001 Jan 21, 2026

Uh oh!

mirgee Jan 21, 2026

Uh oh!

mirgee commented Jan 22, 2026 •

edited

Loading

Uh oh!

matkt left a comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

Conversation

mirgee commented Jan 13, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

matkt commented Jan 19, 2026

Uh oh!

matthew1001 Jan 21, 2026

Choose a reason for hiding this comment

Uh oh!

mirgee Jan 21, 2026

Choose a reason for hiding this comment

Uh oh!

mirgee commented Jan 22, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

matkt left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

mirgee commented Jan 13, 2026 •

edited

Loading

mirgee commented Jan 22, 2026 •

edited

Loading