Requirements for the Application

Formal Requirements

Consensus Connection Requirements

This section specifies what CometBFT expects from the Application. It is structured as a set of formal requirements that can be used for testing and verification of the Application’s logic.

Let p and q be two correct processes. Let rp (resp. rq) be a round of height h where p (resp. q) is the proposer. Let sp,h-1 be p’s Application’s state committed for height h-1. Let vp (resp. vq) be the block that p’s (resp. q’s) CometBFT passes on to the Application via RequestPrepareProposal as proposer of round rp (resp rq), height h, also known as the raw proposal. Let up (resp. uq) the possibly modified block p’s (resp. q’s) Application returns via ResponsePrepareProposal to CometBFT, also known as the prepared proposal.

Process p’s prepared proposal can differ in two different rounds where p is the proposer.

Full execution of blocks at PrepareProposal time stands on CometBFT’s critical path. Thus, Requirement 1 ensures the Application or operator will set a value for TimeoutPropose such that the time it takes to fully execute blocks in PrepareProposal does not interfere with CometBFT’s propose timer. Note that violation of Requirement 1 may just lead to further rounds, but will not compromise liveness.

Busy blockchains might seek to maximize the amount of transactions included in each block. Under those conditions, CometBFT might choose to increase the transactions passed to the Application via RequestPrepareProposal.txs beyond the RequestPrepareProposal.max_tx_bytes limit. The idea is that, if the Application drops some of those transactions, it can still return a transaction list whose byte size is as close to RequestPrepareProposal.max_tx_bytes as possible. Thus, Requirement 2 ensures that the size in bytes of the transaction list returned by the application will never cause the resulting block to go beyond its byte size limit.

Requirement 3 makes sure that blocks proposed by correct processes always pass the correct receiving process’s ProcessProposal check. On the other hand, if there is a deterministic bug in PrepareProposal or ProcessProposal (or in both), strictly speaking, this makes all processes that hit the bug byzantine. This is a problem in practice, as very often validators are running the Application from the same codebase, so potentially all would likely hit the bug at the same time. This would result in most (or all) processes prevoting nil, with the serious consequences on CometBFT’s liveness that this entails. Due to its criticality, Requirement 3 is a target for extensive testing and automated verification.

Requirements 4 and 5 ensure that all correct processes will react in the same way to a proposed block, even if the proposer is Byzantine. However, ProcessProposal may contain a bug that renders the acceptance or rejection of the block non-deterministic, and therefore prevents processes hitting the bug from fulfilling Requirements 4 or 5 (effectively making those processes Byzantine). In such a scenario, CometBFT’s liveness cannot be guaranteed. Again, this is a problem in practice if most validators are running the same software, as they are likely to hit the bug at the same point. There is currently no clear solution to help with this situation, so the Application designers/implementors must proceed very carefully with the logic/implementation of ProcessProposal. As a general rule ProcessProposal SHOULD always accept the block.

According to the Tendermint consensus algorithm, currently adopted in CometBFT, a correct process can broadcast at most one precommit message in round r, height h.

The call to correct process p’s BeginBlock - DeliverTx - EndBlock sequence at height h, with block vp,h passed as parameter, creates state sp,h. Additionally, p’s DeliverTx on transactions creates a set of transaction results Tp,h.

Note that Requirements 11 and 12, combined with the Agreement property of consensus ensure state machine replication, i.e., the Application state evolves consistently at all correct processes.

Also, notice that PrepareProposal has determinism-related requirements associated. Indeed, PrepareProposal is not required to be deterministic:

Mempool Connection Requirements

Let CheckTxCodestx,p,h denote the set of result codes returned by p’s Application, via ResponseCheckTx, to successive calls to RequestCheckTx occurring while the Application is at height h and having transaction tx as parameter. CheckTxCodestx,p,h is a set since p’s Application may return different result codes during height h. If CheckTxCodestx,p,h is a singleton set, i.e. the Application always returned the same result code in ResponseCheckTx while at height h, we define CheckTxCodetx,p,h as the singleton value of CheckTxCodestx,p,h. If CheckTxCodestx,p,h is not a singleton set, CheckTxCodetx,p,h is undefined. Let predicate OK(CheckTxCodetx,p,h) denote whether CheckTxCodetx,p,h is SUCCESS.

Requirement 13 ensures that a transaction will eventually stop oscillating between CheckTx success and failure if it stays in p’s mempool for long enough. This condition on the Application’s behavior allows the mempool to ensure that a transaction will leave the mempool of all full nodes, either because it is expunged everywhere due to failing CheckTx calls, or because it stays valid long enough to be gossipped, proposed and decided. Although Requirement 13 defines a global hstable, application developers can consider such stabilization height as local to process p (hp,stable), without loss for generality. In contrast, the value of b MUST be the same across all processes.

Connection State

CometBFT maintains four concurrent ABCI++ connections, namely Consensus Connection, Mempool Connection, Info/Query Connection, and Snapshot Connection. It is common for an application to maintain a distinct copy of the state for each connection, which are synchronized upon Commit calls.


In principle, each of the four ABCI++ connections operates concurrently with one another. This means applications need to ensure access to state is thread safe. Both the default in-process ABCI client and the default Go ABCI server use a global lock to guard the handling of events across all connections, so they are not concurrent at all. This means whether your app is compiled in-process with CometBFT using the NewLocalClient, or run out-of-process using the SocketServer, ABCI messages from all connections are received in sequence, one at a time.

The existence of this global mutex means Go application developers can get thread safety for application state by routing all reads and writes through the ABCI system. Thus it may be unsafe to expose application state directly to an RPC interface, and unless explicit measures are taken, all queries should be routed through the ABCI Query method.

BeginBlock - DeliverTx - EndBlock

When the consensus algorithm decides on a block, CometBFT uses the sequence of calls to BeginBlock, DeliverTx and EndBlock to send the decided block’s data to the Application, which uses it to transition its state.

The sequence of DeliverTx calls is asynchronous but all those calls are enclosed by calls to BeginBlock and EndBlock which are synchronous.

The Application must remember the latest height from which it has run a successful Commit so that it can tell CometBFT where to pick up from when it recovers from a crash. See information on the Handshake here.


The Application should persist its state during Commit, before returning from it.

Before invoking Commit, CometBFT locks the mempool and flushes the mempool connection. This ensures that no new messages will be received on the mempool connection during this processing step, providing an opportunity to safely update all four connection states to the latest committed state at the same time.

When Commit returns, CometBFT unlocks the mempool.

WARNING: if the ABCI app logic processing the Commit message sends a /broadcast_tx_sync or /broadcast_tx and waits for the response before proceeding, it will deadlock. Executing broadcast_tx calls involves acquiring the mempool lock that CometBFT holds during the Commit call. Synchronous mempool-related calls must be avoided as part of the sequential logic of the Commit function.

Candidate States

CometBFT calls PrepareProposal when it is about to send a proposed block to the network. Likewise, CometBFT calls ProcessProposal upon reception of a proposed block from the network. The proposed block’s data that is disclosed to the Application by these two methods is the following:

The Application may decide to immediately execute the given block (i.e., upon PrepareProposal or ProcessProposal). There are two main reasons why the Application may want to do this:

PrepareProposal/ProcessProposal can be called many times for a given height. Moreover, it is not possible to accurately predict which of the blocks proposed in a height will be decided, being delivered to the Application in that height’s block execution calls. Therefore, the state resulting from executing a proposed block, denoted a candidate state, should be kept in memory as a possible final state for that height. When the block execution functions are called, the Application should check if the decided block corresponds to one of its candidate states; if so, it will apply it as its ExecuteTxState (see Consensus Connection below), which will be persisted during the upcoming Commit call.

Under adverse conditions (e.g., network instability), the consensus algorithm might take many rounds. In this case, potentially many proposed blocks will be disclosed to the Application for a given height. By the nature of Tendermint consensus algorithm, currently adopted in CometBFT, the number of proposed blocks received by the Application for a particular height cannot be bound, so Application developers must act with care and use mechanisms to bound memory usage. As a general rule, the Application should be ready to discard candidate states before block execution, even if one of them might end up corresponding to the decided block and thus have to be reexecuted upon BeginBlock-DeliverTx-EndBlock.

States and ABCI++ Connections

Consensus Connection

The Consensus Connection should maintain an ExecuteTxState — the working state for block execution. It should be updated by the calls to the block execution functions during block execution and committed to disk as the “latest committed state” during Commit. Execution of a proposed block (via PrepareProposal/ProcessProposal) must not update the ExecuteTxState, but rather be kept as a separate candidate state until BeginBlock-DeliverTx-EndBlock confirms which of the candidate states (if any) can be used to update ExecuteTxState.

Mempool Connection

The mempool Connection maintains CheckTxState. CometBFT sequentially processes an incoming transaction (via RPC from client or P2P from the gossip layer) against CheckTxState. If the processing does not return any error, the transaction is accepted into the mempool and CometBFT starts gossipping it. CheckTxState should be reset to the latest committed state at the end of every Commit.

During the execution of a consensus instance, the CheckTxState may be updated concurrently with the ExecuteTxState, as messages may be sent concurrently on the Consensus and Mempool connections. At the end of the consensus instance, as described above, CometBFT locks the mempool and flushes the mempool connection before calling Commit. This ensures that all pending CheckTx calls are responded to and no new ones can begin.

After the Commit call returns, while still holding the mempool lock, CheckTx is run again on all transactions that remain in the node’s local mempool after filtering those included in the block. Parameter Type in RequestCheckTx indicates whether an incoming transaction is new (CheckTxType_New), or a recheck (CheckTxType_Recheck).

Finally, after re-checking transactions in the mempool, CometBFT will unlock the mempool connection. New transactions are once again able to be processed through CheckTx.

Note that CheckTx is just a weak filter to keep invalid transactions out of the mempool and, utimately, ouf of the blockchain. Since the transaction cannot be guaranteed to be checked against the exact same state as it will be executed as part of a (potential) decided block, CheckTx shouldn’t check everything that affects the transaction’s validity, in particular those checks whose validity may depend on transaction ordering. CheckTx is weak because a Byzantine node need not care about CheckTx; it can propose a block full of invalid transactions if it wants. The mechanism ABCI++ has in place for dealing with such behavior is ProcessProposal.

Replay Protection

It is possible for old transactions to be sent again to the Application. This is typically undesirable for all transactions, except for a generally small subset of them which are idempotent.

The mempool has a mechanism to prevent duplicated transactions from being processed. This mechanism is nevertheless best-effort (currently based on the indexer) and does not provide any guarantee of non duplication. It is thus up to the Application to implement an application-specific replay protection mechanism with strong guarantees as part of the logic in CheckTx.

Info/Query Connection

The Info (or Query) Connection should maintain a QueryState. This connection has two purposes: 1) having the application answer the queries CometBFT receives from users (see section Query), and 2) synchronizing CometBFT and the Application at start up time (see Crash Recovery) or after state sync (see State Sync).

QueryState is a read-only copy of ExecuteTxState as it was after the last Commit, i.e. after the full block has been processed and the state committed to disk.

Snapshot Connection

The Snapshot Connection is used to serve state sync snapshots for other nodes and/or restore state sync snapshots to a local node being bootstrapped. Snapshop management is optional: an Application may choose not to implement it.

For more information, see Section State Sync.

Transaction Results

For each transaction within a block, the Application is expected to return a result within ResponseDeliverTx. This section discusses the fields inside ResponseDeliverTx along with the fields in ResponseCheckTx, whose semantics are similar.

The Info and Log fields are non-deterministic values for debugging/convenience purposes. CometBFT logs them but they are otherwise ignored.


Ethereum introduced the notion of gas as an abstract representation of the cost of the resources consumed by nodes when processing a transaction. Every operation in the Ethereum Virtual Machine uses some amount of gas. Gas has a market-variable price based on which miners can accept or reject to execute a particular operation.

Users propose a maximum amount of gas for their transaction; if the transaction uses less, they get the difference credited back. CometBFT adopts a similar abstraction, though uses it only optionally and weakly, allowing applications to define their own sense of the cost of execution.

In CometBFT, the ConsensusParams.Block.MaxGas limits the amount of total gas that can be used by all transactions in a block. The default value is -1, which means the block gas limit is not enforced, or that the concept of gas is meaningless.

Responses contain a GasWanted and GasUsed field. The former is the maximum amount of gas the sender of a transaction is willing to use, and the latter is how much it actually used. Applications should enforce that GasUsed <= GasWanted — i.e. transaction execution or validation should fail before it can use more resources than it requested.

When MaxGas > -1, CometBFT enforces the following rules:

If MaxGas == -1, no rules about gas are enforced.

In v0.34.x and earlier versions, CometBFT does not enforce anything about Gas in consensus, only in the mempool. This means it does not guarantee that committed blocks satisfy these rules. It is the application’s responsibility to return non-zero response codes when gas limits are exceeded when executing the transactions of a block. Since the introduction of PrepareProposal and ProcessProposal in v.0.37.x, it is now possible for the Application to enforce that all blocks proposed (and voted for) in consensus — and thus all blocks decided — respect the MaxGas limits described above.

Since the Application should enforce that GasUsed <= GasWanted when executing a transaction, and it can use PrepareProposal and ProcessProposal to enforce that (sum of GasWanted in a block) <= MaxGas in all proposed or prevoted blocks, we have:

The GasUsed field is ignored by CometBFT.

Specifics of ResponseCheckTx

If Code != 0, it will be rejected from the mempool and hence not broadcasted to other peers and not included in a proposal block.

Data contains the result of the CheckTx transaction execution, if any. It does not need to be deterministic since, given a transaction, nodes’ Applications might have a different CheckTxState values when they receive it and check their validity via CheckTx. CometBFT ignores this value in ResponseCheckTx.

From v0.34.x on, there is a Priority field in ResponseCheckTx that can be used to explicitly prioritize transactions in the mempool for inclusion in a block proposal.

Specifics of ResponseDeliverTx

The BeginBlock-DeliverTx-EndBlock sequence is the workhorse of the blockchain. A sequence of DeliverTx calls delivers the decided block, one transaction at a time, to the Application. The block delivered (and thus the transaction order) is the same at all correct nodes as guaranteed by the Agreement property of consensus.

The Data field contains an array of bytes with the transaction result. It must be deterministic (i.e., the same value must be returned at all nodes), but it can contain arbitrary data. Likewise, the value of Code must be deterministic. If Code != 0, the transaction will be marked invalid, though it is still included in the block. Invalid transactions are not indexed, as they are considered analogous to those that failed CheckTx.

Both the Code and Data are included in a structure that is hashed into the LastResultsHash of the block header in the next height.

Events include any events for the execution, which CometBFT will use to index the transaction by. This allows transactions to be queried according to what events took place during their execution.

Updating the Validator Set

The application may set the validator set during InitChain, and may update it during EndBlock. In both cases, a structure of type ValidatorUpdate is returned.

The InitChain method, used to initialize the Application, can return a list of validators. If the list is empty, CometBFT will use the validators loaded from the genesis file. If the list returned by InitChain is not empty, CometBFT will use its contents as the validator set. This way the application can set the initial validator set for the blockchain.

Applications must ensure that a single set of validator updates does not contain duplicates, i.e. a given public key can only appear once within a given update. If an update includes duplicates, the block execution will fail irrecoverably.

Structure ValidatorUpdate contains a public key, which is used to identify the validator: The public key currently supports three types:

Structure ValidatorUpdate also contains an ìnt64 field denoting the validator’s new power. Applications must ensure that ValidatorUpdate structures abide by the following rules:

Note the updates returned after processing the block at height H will only take effect at block H+2 (see Section Methods).

Consensus Parameters

ConsensusParams are global parameters that apply to all validators in a blockchain. They enforce certain limits in the blockchain, like the maximum size of blocks, amount of gas used in a block, and the maximum acceptable age of evidence. They can be set in InitChain, and updated in EndBlock. These parameters are deterministically set and/or updated by the Application, so all full nodes have the same value at a given height.

List of Parameters

These are the current consensus parameters (as of v0.37.x):

  1. BlockParams.MaxBytes
  2. BlockParams.MaxGas
  3. EvidenceParams.MaxAgeDuration
  4. EvidenceParams.MaxAgeNumBlocks
  5. EvidenceParams.MaxBytes
  6. ValidatorParams.PubKeyTypes
  7. VersionParams.App <!–
  8. SynchronyParams.MessageDelay
  9. SynchronyParams.Precision
  10. TimeoutParams.Propose
  11. TimeoutParams.ProposeDelta
  12. TimeoutParams.Vote
  13. TimeoutParams.VoteDelta
  14. TimeoutParams.Commit
  15. TimeoutParams.BypassCommitTimeout –>

The maximum size of a complete Protobuf encoded block. This is enforced by the consensus algorithm.

This implies a maximum transaction size that is MaxBytes, less the expected size of the header, the validator set, and any included evidence in the block.

The Application should be aware that honest validators may produce and broadcast blocks with up to the configured MaxBytes size. As a result, the consensus timeout parameters adopted by nodes should be configured so as to account for the worst-case latency for the delivery of a full block with MaxBytes size to all validators.

Must have 0 < MaxBytes <= 100 MB.

Bear in mind that the default value for the BlockParams.MaxBytes consensus parameter accepts as valid blocks with size up to 21 MB. If the Application’s use case does not need blocks of that size, or if the impact (specially on bandwidth consumption and block latency) of propagating blocks of that size was not evaluated, it is strongly recommended to wind down this default value.


The maximum of the sum of GasWanted that will be allowed in a proposed block. This is not enforced by the consensus algorithm. It is left to the Application to enforce (ie. if transactions are included past the limit, they should return non-zero codes). It is used by CometBFT to limit the transactions included in a proposed block.

Must have MaxGas >= -1. If MaxGas == -1, no limit is enforced.


This is the maximum age of evidence in time units. This is enforced by the consensus algorithm.

If a block includes evidence older than this (AND the evidence was created more than MaxAgeNumBlocks ago), the block will be rejected (validators won’t vote for it).

Must have MaxAgeDuration > 0.


This is the maximum age of evidence in blocks. This is enforced by the consensus algorithm.

If a block includes evidence older than this (AND the evidence was created more than MaxAgeDuration ago), the block will be rejected (validators won’t vote for it).

Must have MaxAgeNumBlocks > 0.


This is the maximum size of total evidence in bytes that can be committed to a single block. It should fall comfortably under the max block bytes.

Its value must not exceed the size of a block minus its overhead ( ~ BlockParams.MaxBytes).

Must have MaxBytes > 0.


The parameter restricts the type of keys validators can use. The parameter uses ABCI pubkey naming, not Amino names.


This is the version of the ABCI application. <!–


This sets a bound on how long a proposal message may take to reach all validators on a network and still be considered valid.

This parameter is part of the proposer-based timestamps (PBTS) algorithm.


This sets a bound on how skewed a proposer’s clock may be from any validator on the network while still producing valid proposals.

This parameter is part of the proposer-based timestamps (PBTS) algorithm.


Timeout in ms of the propose step of Tendermint consensus algorithm. This value is the initial timeout at every height (round 0).

The value in subsequent rounds is modified by parameter ProposeDelta. When a new height is started, the Propose timeout value is reset to this parameter.

If a node waiting for a proposal message does not receive one matching its current height and round before this timeout, the node will issue a nil prevote for the round and advance to the next step.


Increment in ms to be added to the Propose timeout every time Tendermint consensus algorithm advances one round in a given height.

When a new height is started, the Propose timeout value is reset.


Timeout in ms of the prevote and precommit steps of Tendermint consensus algorithm. This value is the initial timeout at every height (round 0).

The value in subsequent rounds is modified by parameter VoteDelta. When a new height is started, the Vote timeout value is reset to this parameter.

The Vote timeout does not begin until a quorum of votes has been received. Once a quorum of votes has been seen and this timeout elapses, Tendermint will procced to the next step of the consensus algorithm. If Tendermint receives all of the remaining votes before the end of the timeout, it will proceed to the next step immediately.


Increment in ms to be added to the Vote timeout every time Tendermint consensus algorithm advances one round in a given height.

When a new height is started, the Vote timeout value is reset.


This configures how long Tendermint consensus algorithm will wait after receiving a quorum of precommits before beginning consensus for the next height. This can be used to allow slow precommits to arrive for inclusion in the next height before progressing.


This configures the node to proceed immediately to the next height once the node has received all precommits for a block, forgoing the remaining commit timeout. Setting this parameter to false (the default) causes Tendermint to wait for the full commit timeout configured in TimeoutParams.Commit. –> <!–


This parameter is either 0 or a positive height at which vote extensions become mandatory. If the value is zero (which is the default), vote extensions are not required. Otherwise, at all heights greater than the configured height H vote extensions must be present (even if empty). When the configured height H is reached, PrepareProposal will not include vote extensions yet, but ExtendVote and VerifyVoteExtension will be called. Then, when reaching height H+1, PrepareProposal will include the vote extensions from height H. For all heights after H

Must always be set to a future height. Once set to a value different from 0, its value must not be changed. –>

Updating Consensus Parameters

The application may set the ConsensusParams during InitChain, and update them during EndBlock. If the ConsensusParams is empty, it will be ignored. Each field that is not empty will be applied in full. For instance, if updating the Block.MaxBytes, applications must also set the other Block fields (like Block.MaxGas), even if they are unchanged, as they will otherwise cause the value to be updated to the default.


ResponseInitChain includes a ConsensusParams parameter. If ConsensusParams is nil, CometBFT will use the params loaded in the genesis file. If ConsensusParams is not nil, CometBFT will use it. This way the application can determine the initial consensus parameters for the blockchain.

EndBlock, PrepareProposal/ProcessProposal

ResponseEndBlock accepts a ConsensusParams parameter. If ConsensusParams is nil, CometBFT will do nothing. If ConsensusParams is not nil, CometBFT will use it. This way the application can update the consensus parameters over time.

The updates returned in block H will take effect right away for block H+1.


Query is a generic method with lots of flexibility to enable diverse sets of queries on application state. CometBFT makes use of Query to filter new peers based on ID and IP, and exposes Query to the user over RPC.

Note that calls to Query are not replicated across nodes, but rather query the local node’s state - hence they may return stale reads. For reads that require consensus, use a transaction.

The most important use of Query is to return Merkle proofs of the application state at some height that can be used for efficient application-specific light-clients.

Note CometBFT has technically no requirements from the Query message for normal operation - that is, the ABCI app developer need not implement Query functionality if they do not wish to.

Query Proofs

The CometBFT block header includes a number of hashes, each providing an anchor for some type of proof about the blockchain. The ValidatorsHash enables quick verification of the validator set, the DataHash gives quick verification of the transactions included in the block.

The AppHash is unique in that it is application specific, and allows for application-specific Merkle proofs about the state of the application. While some applications keep all relevant state in the transactions themselves (like Bitcoin and its UTXOs), others maintain a separated state that is computed deterministically from transactions, but is not contained directly in the transactions themselves (like Ethereum contracts and accounts). For such applications, the AppHash provides a much more efficient way to verify light-client proofs.

ABCI applications can take advantage of more efficient light-client proofs for their state as follows:

For instance, this allows an application’s light-client to verify proofs of absence in the application state, something which is much less efficient to do using the block hash.

Some applications (eg. Ethereum, Cosmos-SDK) have multiple “levels” of Merkle trees, where the leaves of one tree are the root hashes of others. To support this, and the general variability in Merkle proofs, the ResponseQuery.Proof has some minimal structure:

message ProofOps {
  repeated ProofOp ops = 1

message ProofOp {
  string type = 1;
  bytes key   = 2;
  bytes data  = 3;

Each ProofOp contains a proof for a single key in a single Merkle tree, of the specified type. This allows ABCI to support many different kinds of Merkle trees, encoding formats, and proofs (eg. of presence and absence) just by varying the type. The data contains the actual encoded proof, encoded according to the type. When verifying the full proof, the root hash for one ProofOp is the value being verified for the next ProofOp in the list. The root hash of the final ProofOp in the list should match the AppHash being verified against.

Peer Filtering

When CometBFT connects to a peer, it sends two queries to the ABCI application using the following paths, with no additional data:

If either of these queries return a non-zero ABCI code, CometBFT will refuse to connect to the peer.


Queries are directed at paths, and may optionally include additional data.

The expectation is for there to be some number of high level paths differentiating concerns, like /p2p, /store, and /app. Currently, CometBFT only uses /p2p, for filtering peers. For more advanced use, see the implementation of Query in the Cosmos-SDK.

Crash Recovery

On startup, CometBFT calls the Info method on the Info Connection to get the latest committed state of the app. The app MUST return information consistent with the last block it succesfully completed Commit for.

If the app succesfully committed block H, then last_block_height = H and last_block_app_hash = <hash returned by Commit for block H>. If the app failed during the Commit of block H, then last_block_height = H-1 and last_block_app_hash = <hash returned by Commit for block H-1, which is the hash in the header of block H>.

We now distinguish three heights, and describe how CometBFT syncs itself with the app.

storeBlockHeight = height of the last block CometBFT saw a commit for
stateBlockHeight = height of the last block for which CometBFT completed all
    block processing and saved all ABCI results to disk
appBlockHeight = height of the last block for which ABCI app succesfully
    completed Commit

Note we always have storeBlockHeight >= stateBlockHeight and storeBlockHeight >= appBlockHeight Note also CometBFT never calls Commit on an ABCI app twice for the same height.

The procedure is as follows.

First, some simple start conditions:

If appBlockHeight == 0, then call InitChain.

If storeBlockHeight == 0, we’re done.

Now, some sanity checks:

If storeBlockHeight < appBlockHeight, error If storeBlockHeight < stateBlockHeight, panic If storeBlockHeight > stateBlockHeight+1, panic

Now, the meat:

If storeBlockHeight == stateBlockHeight && appBlockHeight < storeBlockHeight, replay all blocks in full from appBlockHeight to storeBlockHeight. This happens if we completed processing the block, but the app forgot its height.

If storeBlockHeight == stateBlockHeight && appBlockHeight == storeBlockHeight, we’re done. This happens if we crashed at an opportune spot.

If storeBlockHeight == stateBlockHeight+1 This happens if we started processing the block but didn’t finish.

If appBlockHeight < stateBlockHeight replay all blocks in full from appBlockHeight to storeBlockHeight-1, and replay the block at storeBlockHeight using the WAL. This happens if the app forgot the last block it committed.

If appBlockHeight == stateBlockHeight, replay the last block (storeBlockHeight) in full. This happens if we crashed before the app finished Commit

If appBlockHeight == storeBlockHeight update the state using the saved ABCI responses but dont run the block against the real app. This happens if we crashed after the app finished Commit but before CometBFT saved the state.

State Sync

A new node joining the network can simply join consensus at the genesis height and replay all historical blocks until it is caught up. However, for large chains this can take a significant amount of time, often on the order of days or weeks.

State sync is an alternative mechanism for bootstrapping a new node, where it fetches a snapshot of the state machine at a given height and restores it. Depending on the application, this can be several orders of magnitude faster than replaying blocks.

Note that state sync does not currently backfill historical blocks, so the node will have a truncated block history - users are advised to consider the broader network implications of this in terms of block availability and auditability. This functionality may be added in the future.

For details on the specific ABCI calls and types, see the methods section.

Taking Snapshots

Applications that want to support state syncing must take state snapshots at regular intervals. How this is accomplished is entirely up to the application. A snapshot consists of some metadata and a set of binary chunks in an arbitrary format:

For a snapshot to be considered the same across nodes, all of these fields must be identical. When sent across the network, snapshot metadata messages are limited to 4 MB.

When a new node is running state sync and discovering snapshots, CometBFT will query an existing application via the ABCI ListSnapshots method to discover available snapshots, and load binary snapshot chunks via LoadSnapshotChunk. The application is free to choose how to implement this and which formats to use, but must provide the following guarantees:

A very basic approach might be to use a datastore with MVCC transactions (such as RocksDB), start a transaction immediately after block commit, and spawn a new thread which is passed the transaction handle. This thread can then export all data items, serialize them using e.g. Protobuf, hash the byte stream, split it into chunks, and store the chunks in the file system along with some metadata - all while the blockchain is applying new blocks in parallel.

A more advanced approach might include incremental verification of individual chunks against the chain app hash, parallel or batched exports, compression, and so on.

Old snapshots should be removed after some time - generally only the last two snapshots are needed (to prevent the last one from being removed while a node is restoring it).

Bootstrapping a Node

An empty node can be state synced by setting the configuration option statesync.enabled = true. The node also needs the chain genesis file for basic chain info, and configuration for light client verification of the restored snapshot: a set of CometBFT RPC servers, and a trusted header hash and corresponding height from a trusted source, via the statesync configuration section.

Once started, the node will connect to the P2P network and begin discovering snapshots. These will be offered to the local application via the OfferSnapshot ABCI method. Once a snapshot is accepted CometBFT will fetch and apply the snapshot chunks. After all chunks have been successfully applied, CometBFT verifies the app’s AppHash against the chain using the light client, then switches the node to normal consensus operation.

Snapshot Discovery

When the empty node joins the P2P network, it asks all peers to report snapshots via the ListSnapshots ABCI call (limited to 10 per node). After some time, the node picks the most suitable snapshot (generally prioritized by height, format, and number of peers), and offers it to the application via OfferSnapshot. The application can choose a number of responses, including accepting or rejecting it, rejecting the offered format, rejecting the peer who sent it, and so on. CometBFT will keep discovering and offering snapshots until one is accepted or the application aborts.

Snapshot Restoration

Once a snapshot has been accepted via OfferSnapshot, CometBFT begins downloading chunks from any peers that have the same snapshot (i.e. that have identical metadata fields). Chunks are spooled in a temporary directory, and then given to the application in sequential order via ApplySnapshotChunk until all chunks have been accepted.

The method for restoring snapshot chunks is entirely up to the application.

During restoration, the application can respond to ApplySnapshotChunk with instructions for how to continue. This will typically be to accept the chunk and await the next one, but it can also ask for chunks to be refetched (either the current one or any number of previous ones), P2P peers to be banned, snapshots to be rejected or retried, and a number of other responses - see the ABCI reference for details.

If CometBFT fails to fetch a chunk after some time, it will reject the snapshot and try a different one via OfferSnapshot - the application can choose whether it wants to support restarting restoration, or simply abort with an error.

Snapshot Verification

Once all chunks have been accepted, CometBFT issues an Info ABCI call to retrieve the LastBlockAppHash. This is compared with the trusted app hash from the chain, retrieved and verified using the light client. CometBFT also checks that LastBlockHeight corresponds to the height of the snapshot.

This verification ensures that an application is valid before joining the network. However, the snapshot restoration may take a long time to complete, so applications may want to employ additional verification during the restore to detect failures early. This might e.g. include incremental verification of each chunk against the app hash (using bundled Merkle proofs), checksums to protect against data corruption by the disk or network, and so on. However, it is important to note that the only trusted information available is the app hash, and all other snapshot metadata can be spoofed by adversaries.

Apps may also want to consider state sync denial-of-service vectors, where adversaries provide invalid or harmful snapshots to prevent nodes from joining the network. The application can counteract this by asking CometBFT to ban peers. As a last resort, node operators can use P2P configuration options to whitelist a set of trusted peers that can provide valid snapshots.

Transition to Consensus

Once the snapshots have all been restored, CometBFT gathers additional information necessary for bootstrapping the node (e.g. chain ID, consensus parameters, validator sets, and block headers) from the genesis file and light client RPC servers. It also calls Info to verify the following:

Once the state machine has been restored and CometBFT has gathered this additional information, it transitions to block sync (if enabled) to fetch any remaining blocks up the chain head, and then transitions to regular consensus operation. At this point the node operates like any other node, apart from having a truncated block history at the height of the restored snapshot.

Decorative Orb