- Introduction
- Use cases
- Architecture
- Required Setup
- Integrations currently live
- Binary Installation
- Run a node
- Creating a pool with the runtime
This runtime validates and archives blocks from any tendermint based blockchain application. It only stores blocks from and to a given height and makes them available to directly download them from the storage provider, or directly sync them into the blockchain nodes itself, therefore enabling block sync via validated KYVE blocks.
Since storage pools which use this runtime archive validated and historical blocks blockchain nodes can use those archived blocks to bootstrap themselves and sync to the current network height. This may make expensive archival nodes obsolete since those blocks are already permanently and immutably archived. Additionally, block data can be retrieved over an ELT pipeline, further analyzing and using it for different applications like block explorers.
This section explains how Tendermint blocks are collected for block-sync (fast sync) and archived.
graph TB
subgraph "Data Source"
TM[Tendermint Node<br/>Cosmos SDK Chain]
end
subgraph "KYVE Protocol Node"
Runtime[Tendermint BSync Runtime]
Cache[Local Cache]
Protocol[Protocol Core]
end
subgraph "Storage & Chain"
Storage[Storage Provider<br/>Arweave]
Chain[KYVE Chain]
end
TM -->|1. /block?height=N<br/>fetch block only| Runtime
Runtime -->|2. Prevalidate<br/>height & chain_id| Runtime
Runtime -->|3. No transform<br/>pass through| Runtime
Runtime -->|4. Cache| Cache
Cache -->|5. Bundle 100 blocks| Protocol
Protocol -->|6. Compress| Protocol
Protocol -->|7. Upload| Storage
Storage -->|8. Propose| Chain
Chain -->|9. Validators download| Storage
Storage -->|10. Validate| Runtime
Runtime -->|11. Vote VALID/INVALID| Chain
style Runtime fill:#e1f5ff
style Protocol fill:#fff4e1
style Chain fill:#ffe1e1
Key Differences from Standard Tendermint Runtime:
-
Block Data Only: Fetches only block data, not
block_results- Significantly faster data collection
- Smaller bundle sizes (~50% reduction)
- Ideal for fast sync use cases where events aren't needed
-
No Transformation: Block data is already deterministic
- No event sorting required
- No attribute manipulation needed
- Simpler and faster validation
-
Exact Validation: Simple JSON comparison
- No ABSTAIN votes needed
- No non-determinism handling required
- Either VALID or INVALID
-
Optimized for Speed: Designed for rapid blockchain synchronization
- Nodes can quickly download validated blocks
- Fast sync from genesis to current height
- Minimal overhead compared to running full archival node
The runtime implements a streamlined version of the IRuntime interface:
Fetches block at a given height:
- Calls
/block?height={key}to get block data - Returns only the
blockobject (not block_results) - Significantly faster than full block + block_results fetch
Data Structure:
{
key: "block_height",
value: {
header: {
height: "block_height",
chain_id: "cosmoshub-4",
time: "2021-01-01T00:00:00Z",
// ... other header fields
},
data: {
txs: [/* base64 encoded transactions */]
},
evidence: { /* evidence of misbehavior */ },
last_commit: { /* commit signatures */ }
}
}Validates basic block properties:
- Block exists: Ensures value is defined
- Height matches: Verifies block height matches key
- Chain ID matches: Confirms block belongs to configured network
No schema validation is performed (unlike standard Tendermint runtime) - relying on Tendermint node to provide valid blocks.
No transformation applied:
- Block data from Tendermint is already deterministic
- No event sorting needed (no events included)
- Data passed through as-is
Simple exact JSON comparison:
- Compares proposed and validation items with
JSON.stringify - Returns VALID if exact match
- Returns INVALID if any difference
No ABSTAIN votes - block data is fully deterministic, so validators always agree or disagree.
Creates merkle root from bundle:
- Hashes all block data in the bundle
- Generates merkle tree from hashes
- Returns merkle root as hex string
- Used for compact on-chain verification
Increments block height by 1 to get next block.
Comparison with Standard Tendermint Runtime:
| Feature | tendermint-bsync | tendermint |
|---|---|---|
| Data Fetched | Block only | Block + block_results |
| Transform | None | Sort events, remove logs |
| Validation | Exact match | Exact match → ABSTAIN if events differ |
| Vote Types | VALID, INVALID | VALID, INVALID, ABSTAIN |
| Use Case | Fast sync | Full data archive + event indexing |
| Bundle Size | ~50% smaller | Larger (includes events) |
| Speed | Faster | Slower (more data to process) |
When to Use tendermint-bsync:
- Fast blockchain synchronization
- Block Explorer data feeds (without events)
- Historical block archival for sync purposes
- When event data is not required
When to Use standard tendermint:
- Need transaction events for indexing
- Building analytics on blockchain events
- Full blockchain data archival
- When complete block execution results are required
This runtime requires the node operator to run a tendermint node which is used as the source and the KYVE protocol node. Depending on which tendermint chain gets archived the minimum hardware requirements are at least the min requirements of that tendermint node.
The following integrations are running on this runtime and are currently live.
(planned - currently in test)
- Pool ID: 0
- Chain ID: cosmoshub-4
- Base Height: 5200791
- Cosmos Hub // cosmoshub-4
- Pool ID: 24
- Chain ID: cosmoshub-4
- Base Height: 5200791
This section explains how to install a protocol node with this runtime. This is only relevant for protocol node operators who want to run a node in a pool which has this runtime.
The first option to install the binary is to build it from source. For that you have to execute the following commands:
git clone git@github.com:KYVENetwork/kyvejs.git
cd kyvejsIf you want to build a specific version you can checkout the tag and continue from the version branch. If you want to build the latest version you can skip this step.
git checkout tags/@kyvejs/tendermint-bsync@x.x.x -b x.x.xAfter you have cloned the project and have the desired version the dependencies can be installed and the project build:
yarn install
yarn setupFinally, you can build the runtime binaries.
INFO: During the binary build log warnings can occur. You can safely ignore them.
cd integrations/tendermint-bsync
yarn build:binariesYou can verify the installation with printing the version:
./out/kyve-linux-64 versionAfter the build succeeded you can find the binaries in the out folder where you can move them to use
desired location (like KYSOR).
You can find all prebuilt binaries in the releases of the kyvejs repository. For this specific runtime they can be found here.
You can verify the installation with printing the version:
./kyve-linux-64 versionOnce you have downloaded the binary for the correct platform and version you can simply unzip them and move them to your desired location (like KYSOR).
This section explains which runtime specific setup you must have in order to run a node. This is only relevant for protocol node operators who have already installed the binary (previous section) and want to run a node in a pool which has this runtime.
Depending on the integration which are currently live the following setup has to be done.
INFO: This pool archives the current Cosmos Hub (Chain Id: cosmoshub-4) from block 5,200,791 to the current block (~14M).
Every protocol node runner will run their own Cosmos Hub blockchain node as a data source. This ensures that the data which is getting proposed and validated actually comes from decentralized sources. Furthermore, since the gaia blockchain node only serves valid blocks we further increase the validation of this data. With that setup a user who wants to join this pool first has to sync his gaia node to the current height the pool has already archived the blocks and then start the actual KYVE protocol node.
This architecture diagram summarizes the setup of the Cosmos Hub integration on KYVE:
Here this runtime is responsible for communicating with the tendermint application (purple) - in this case gaia, and forwarding the data to the KYVE core protocol. The KYVE core then handles the communication with the pool. This entire process (yellow) is the KYVE protocol node. The resulting data are the blocks from the tendermint application - validated and permanently stored on a storage provider like Arweave.
The following minimum hardware requirements have to be met:
- RAM: 16GB
- Storage: 512GB*
*This can increase over time
The first step is to start an archival gaia node. For that the gaia binary with the version v4.2.1 has
to be installed. You can follow the official installation instructions here or download the binary directly from here.
You can verify the successful installation with
./gaiad version
4.2.1
In order to setup the gaia config first choose a moniker and init gaia:
./gaiad init <your-moniker> --chain-id cosmoshub-4To install the genesis file execute the following command:
wget https://raw.githubusercontent.com/cosmos/mainnet/master/genesis/genesis.cosmoshub-4.json.gz
gzip -d genesis.cosmoshub-4.json.gz
mv genesis.cosmoshub-4.json ~/.gaia/config/genesis.jsonPeers can be added via this addrbook which can be retrieved here:
wget https://dl2.quicksync.io/json/addrbook.cosmos.json
mv addrbook.cosmos.json ~/.gaia/config/addrbook.json
chmod 666 ~/.gaia/config/addrbook.jsonTIP: You can also add persistent_peers from Polkachu to ensure that you will actually find peers where you can sync with the network: https://polkachu.com/live_peers/cosmos
For pruning the following settings are recommended to decrease the disk usage:
~/.gaia/config/config.toml
[tx_index]
indexer = "null"~/.gaia/config/app.toml
pruning = "everything"
index-events = [""]Finally, the node can be started:
NOTE: For some users it might be required to increase the limit of open files on your system with ulimit -n 65536
INFO: Since the genesis file is quite big (over 100MB) the starting process can take a serveral minutes before the node starts to sync blocks.
./gaiad start --x-crisis-skip-assert-invariantsNow you have to sync blocks until the latest summary of the pool is reached. For example if the latest pool summary is 6,000,000 you can check if the node has synced the blocks until that height with:
curl http://localhost:26657/block?height=6000000If it returns a valid block response you can continue with starting the actual KYVE protocol node and start participating in the validation and archival process.
TIP: to save storage space you can start pruning blocks manually after they have been archived by the pool since after that they are not needed anymore.
TIP: to make it easier during chain upgrades we also recommend running the chain with cosmosvisor. You can find official instructions here
The remaining installation of the KYVE protocol node is the same for every other protocol node. You can now follow the official docs starting from here
The Binaries of this runtime with which to join the pool can be found here: https://github.com/KYVENetwork/kyvejs/releases?q=tendermint-bsync
This section explains how you can create a storage pool on KYVE with this specific runtime. This is only relevant for users or projects, who are interested in archiving and validating a new data source.
This runtime requires the following config format in order to run:
{
"network": "CHAIN_ID_OF_INTEGRATION",
"rpc": "https://rpc-endpoint-of-integration:26657"
}Here the properties have the following reason:
network: the chain ID of the network. This is a check to verify that only blocks from this network are validated and archived. The runtime rejects blocks that do not match with this IDrpc: the default rpc endpoint of the network to collect blocks from. This should only be a base URL without a trailing slash. This can be a public rpc endpoint from a dedicated provider or an URL pointing to localhost in order to signal that every protocol node has to host their own blockchain node
This config should then be stringified on the pool and should look like this:
{
"config": "{\"network\":\"CHAIN_ID_OF_INTEGRATION\",\"rpc\":\"https://rpc-endpoint-of-integration:26657\"}"
}You can override the pool's RPC endpoint using the KYVEJS_TENDERMINT_BSYNC_RPC environment variable. This is useful for:
- Using a local node instead of a remote endpoint
- Testing with different RPC providers
- Running multiple pools with different endpoints
export KYVEJS_TENDERMINT_BSYNC_RPC="https://my-custom-rpc-endpoint:26657"When set, this environment variable takes precedence over the rpc value in the pool config.
In order to create a pool it has to go through the Governance process (more on that can be found here). An example proposal with which a storage pool with this runtime could be created can be found below:
{
"messages": [
{
"@type": "/kyve.pool.v1beta1.MsgCreatePool",
"authority": "kyve10d07y265gmmuvt4z0w9aw880jnsr700jdv7nah",
"name": "<your pool name>",
"runtime": "@kyvejs/tendermint-bsync",
"logo": "ar://<your logo stored on arweave>",
"config": "<your config like described above>",
"start_key": "<the initial height of the tendermint chain>",
"upload_interval": "120", // 120s is the recommended value
"operating_cost": "<your base bundle reward>", // for example 1000000 if the base reward per bundle should be 1 $KYVE
"min_delegation": "<your required min delegation", // for example 1000000000 if the pool should only run if more than 1000 $KYVE are bonded in this pool
"max_bundle_size": "100", // 100 blocks per bundle is the recommended value
"version": "<runtime version>", // the current version of this runtime
"binaries": "{\"kyve-linux-arm64\":\"<linux-arm64 binary download URL>\",\"kyve-linux-x64\":\"<linux-x64 binary download URL>\",\"kyve-macos-x64\":\"<macos-x64 binary download URL>\"}", // download URLs of binaries for KYSOR
"storageProviderId": "1", // Arweave is the recommended storage provider
"compressionId": "1" // Gzip is the recommended bundle compression
}
],
"metadata": "<your ipfs metadata info>", // gov proposal metadata
"deposit": "<your deposit>ukyve" // proposal deposit, check the required min deposit
}Once your proposal is ready you can submit it to the network. Please follow the official governance process to increase the chances of getting your proposal accepted.
