Skip to main content

Join the Cosmos Hub Mainnet

The current Cosmos Hub mainnet, cosmoshub-4, has been performing in place store migration upgrades as of the Delta Upgrade July 2021. The most recent upgrade is Gaia v14.1.x Dec 2023. This type of upgrade preserves the same chain-id but state before the upgrade height is only accessible by corresponding versions of the binary:

Release History

  • use gaia v5.0.x (Delta) for queries of state between height 6,910,000 and 8,695,000
  • use gaia v6.0.x (Vega) between 8,695,000 and 10,085,397
  • use gaia v7.0.x (Theta) between 10,085,397 and 14,099,412
  • use gaia v8.0.x (Rho) between 14,099,412 and 14,470,501
  • use gaia v9.0.x (Lambda) between 14,470,501 and 15,213,800
  • use gaia v9.1.x between 15,213,800 and 15,816,200
  • use gaia v10.0.x between 15,816,200 and 16,596,000
  • use gaia v11.x between 16,596,000 and 16,985,500
  • use gaia v12.x between 16,985,500 and 17,380,000
  • use gaia v13.x between 17,380,000 and 18,262,000
  • use gaia v14.1.x from 18,262,000

For more details, see the history of upgrades or visit the migration section of the Hub's docs.

This guide includes full instructions for joining the mainnet either as an archive/full node or a pruned node.

For instructions to bootstrap a node via Quicksync or State Sync, see the Quickstart Guide

For instructions to join as a validator, please also see the Validator Guide.

Overview

Background

The current Cosmos Hub mainnet cosmoshub-4. Visit the migration section of the Hub's docs for more information on previous chain migrations.

Explorers

There are many explorers for the Cosmos Hub. For reference while setting up a node, here are a few recommendations:

Getting Started

Make sure the following prerequisites are completed:

Hardware

Running a full archive node can be resource intensive as the full current cosmoshub-4 state is over 1.4TB. For those who wish to run state sync or use quicksync, the following hardware configuration is recommended:

Node TypeRAMStorage
Validator32GB500GB-2TB*
Full16GB2TB
Default16GB1TB

* Storage size for validators will depend on level of pruning.

General Configuration

Make sure to walk through the basic setup and configuration. Operators will need to initialize gaiad, download the genesis file for cosmoshub-4, and set persistent peers and/or seeds for startup.

Initialize Chain

Choose a custom moniker for the node and initialize. By default, the init command creates the ~/.gaia directory with subfolders config and data. In the /config directory, the most important files for configuration are app.toml and config.toml.

gaiad init <custom-moniker>

Note: Monikers can contain only ASCII characters. Using Unicode characters is not supported and renders the node unreachable.

The moniker can be edited in the ~/.gaia/config/config.toml file:

# A custom human readable name for this node
moniker = "<custom_moniker>"

Genesis File

Once the node is initialized, download the genesis file and move to the /config directory of the Gaia home directory.

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.json

Seeds & Peers

Upon startup the node will need to connect to peers. If there are specific nodes a node operator is interested in setting as seeds or as persistent peers, this can be configured in ~/.gaia/config/config.toml

# Comma separated list of seed nodes to connect to
seeds = "<seed node id 1>@<seed node address 1>:26656,<seed node id 2>@<seed node address 2>:26656"

# Comma separated list of nodes to keep persistent connections to
persistent_peers = "<node id 1>@<node address 1>:26656,<node id 2>@<node address 2>:26656"

Node operators can optionally download the Quicksync address book. Make sure to move this to ~/.gaia/config/addrbook.json.

Gas & Fees

On Cosmos Hub mainnet, the accepted denom is uatom, where 1atom = 1.000.000uatom

Transactions on the Cosmos Hub network need to include a transaction fee in order to be processed. This fee pays for the gas required to run the transaction. The formula is the following:

fees = ceil(gas * gasPrices)

Gas is the smallest unit or pricing value required to perform a transaction. Different transactions require different amounts of gas. The gas amount for a transaction is calculated as it is being processed, but it can be estimated beforehand by using the auto value for the gas flag. The gas estimate can be adjusted with the flag --gas-adjustment (default 1.0) to ensure enough gas is provided for the transaction.

The gasPrice is the price of each unit of gas. Each validator sets a min-gas-price value, and will only include transactions that have a gasPrice greater than their min-gas-price.

The transaction fees are the product of gas and gasPrice. The higher the gasPrice/fees, the higher the chance that a transaction will get included in a block.

For mainnet, the recommended gas-prices is 0.0025uatom.

A full-node keeps unconfirmed transactions in its mempool. In order to protect it from spam, it is better to set a minimum-gas-prices that the transaction must meet in order to be accepted in the node's mempool. This parameter can be set in ~/.gaia/config/app.toml.

# The minimum gas prices a validator is willing to accept for processing a
# transaction. A transaction's fees must meet the minimum of any denomination
# specified in this config (e.g. 0.25token1;0.0001token2).
minimum-gas-prices = "0.0025uatom"

The initial recommended min-gas-prices is 0.0025uatom, but this can be changed later.

Pruning of State

Note: This is an optional configuration.

There are four strategies for pruning state. These strategies apply only to state and do not apply to block storage. A node operator may want to consider custom pruning if node storage is a concern or there is an interest in running an archive node.

To set pruning, adjust the pruning parameter in the ~/.gaia/config/app.toml file. The following pruning state settings are available:

  1. everything: Prune all saved states other than the current state.
  2. nothing: Save all states and delete nothing.
  3. default: Save the last 100 states and the state of every 10,000th block.
  4. custom: Specify pruning settings with the pruning-keep-recent, pruning-keep-every, and pruning-interval parameters.

By default, every node is in default mode which is the recommended setting for most environments. If a node operator wants to change their node's pruning strategy then this must be done before the node is initialized.

In ~/.gaia/config/app.toml

# default: the last 100 states are kept in addition to every 500th state; pruning at 10 block intervals
# nothing: all historic states will be saved, nothing will be deleted (i.e. archiving node)
# everything: all saved states will be deleted, storing only the current state; pruning at 10 block intervals
# custom: allow pruning options to be manually specified through 'pruning-keep-recent', 'pruning-keep-every', and 'pruning-interval'
pruning = "custom"

# These are applied if and only if the pruning strategy is custom.
pruning-keep-recent = "10"
pruning-keep-every = "1000"
pruning-interval = "10"

Passing a flag when starting gaia will always override settings in the app.toml file. To change the node's pruning setting to everything mode then pass the ---pruning everything flag when running gaiad start.

Note: If running the node with pruned state, it will not be possible to query the heights that are not in the node's store.

REST API

Note: This is an optional configuration.

By default, the REST API is disabled. To enable the REST API, edit the ~/.gaia/config/app.toml file, and set enable to true in the [api] section.

###############################################################################
### API Configuration ###
###############################################################################
[api]
# Enable defines if the API server should be enabled.
enable = true
# Swagger defines if swagger documentation should automatically be registered.
swagger = false
# Address defines the API server to listen on.
address = "tcp://0.0.0.0:1317"

Optionally activate swagger by setting swagger to true or change the port of the REST API in the parameter address. After restarting the application, access the REST API on <NODE IP>:1317.

GRPC

Note: This is an optional configuration.

By default, gRPC is enabled on port 9090. The ~/.gaia/config/app.toml file is where changes can be made in the gRPC section. To disable the gRPC endpoint, set enable to false. To change the port, use the address parameter.

###############################################################################
### gRPC Configuration ###
###############################################################################
[grpc]
# Enable defines if the gRPC server should be enabled.
enable = true
# Address defines the gRPC server address to bind to.
address = "0.0.0.0:9090"

Sync Options

There are three main ways to sync a node on the Cosmos Hub; Blocksync, State Sync, and Quicksync. See the matrix below for the Hub's recommended setup configuration. This guide will focus on syncing two types of common nodes; full and pruned. For further information on syncing to run a validator node, see the section on Validators.

There are two types of concerns when deciding which sync option is right. Data integrity refers to how reliable the data provided by a subset of network participants is. Historical data refers to how robust and inclusive the chain’s history is.

Low Data IntegrityHigh Data Integrity
Minimal Historical DataQuicksync - PrunedState Sync
Moderate Historical DataQuicksync - Default
Full Historical DataQuicksync - ArchiveBlocksync

If a node operator wishes to run a full node, it is possible to start from scratch but will take a significant amount of time to catch up. Node operators not concerned with rebuilding original state from the beginning of cosmoshub-4 can also leverage Quicksync's available archive history.

For operators interested in bootstrapping a pruned node, either Quicksync or State Sync would be sufficient.

Make sure to consult the hardware section for guidance on the best configuration for the type of node operating.

Blocksync

Blocksync is faster than traditional consensus and syncs the chain from genesis by downloading blocks and verifying against the merkle tree of validators. For more information see CometBFT's Fastsync Docs

When syncing via Blocksync, node operators will either need to manually upgrade the chain or set up Cosmovisor to upgrade automatically.

For more information on performing the manual upgrades, see Releases & Upgrades.

It is possible to sync from previous versions of the Cosmos Hub. See the matrix below for the correct gaia version. See the mainnet archive for historical genesis files.

Chain IdGaia Version
cosmoshub-4v4.2.1
cosmoshub-3v2.0.x
cosmoshub-2v1.0.x
cosmoshub-1v0.0.x
Getting Started

Start Gaia to begin syncing with the skip-invariants flag. For more information on this see Verify Mainnet.

gaiad start --x-crisis-skip-assert-invariants

The node will begin rebuilding state until it hits the first upgrade height at block 6910000. If Cosmovisor is set up then there's nothing else to do besides wait, otherwise the node operator will need to perform the manual upgrade twice.

State Sync

State Sync is an efficient and fast way to bootstrap a new node, and it works by replaying larger chunks of application state directly rather than replaying individual blocks or consensus rounds. For more information, see CometBFT's State Sync docs.

To enable state sync, visit an explorer to get a recent block height and corresponding hash. A node operator can choose any height/hash in the current bonding period, but as the recommended snapshot period is 1000 blocks, it is advised to choose something close to current height - 1000.

With the block height and hash selected, update the configuration in ~/.gaia/config/config.toml to set enable = true, and populate the trust_height and trust_hash. Node operators can configure the rpc servers to a preferred provider, but there must be at least two entries. It is important that these are two rpc servers the node operator trusts to verify component parts of the chain state. While not recommended, uniqueness is not currently enforced, so it is possible to duplicate the same server in the list and still sync successfully.

Note: In the future, the RPC server requirement will be deprecated as state sync is moved to the p2p layer in Tendermint 0.38.

#######################################################
### State Sync Configuration Options ###
#######################################################
[statesync]
# State sync rapidly bootstraps a new node by discovering, fetching, and restoring a state machine
# snapshot from peers instead of fetching and replaying historical blocks. Requires some peers in
# the network to take and serve state machine snapshots. State sync is not attempted if the node
# has any local state (LastBlockHeight > 0). The node will have a truncated block history,
# starting from the height of the snapshot.
enable = true

# RPC servers (comma-separated) for light client verification of the synced state machine and
# retrieval of state data for node bootstrapping. Also needs a trusted height and corresponding
# header hash obtained from a trusted source, and a period during which validators can be trusted.
#
# For Cosmos SDK-based chains, trust_period should usually be about 2/3 of the unbonding time (~2
# weeks) during which they can be financially punished (slashed) for misbehavior.
rpc_servers = "https://cosmos-rpc.polkachu.com:443,https://rpc-cosmoshub-ia.cosmosia.notional.ventures:443,https://rpc.cosmos.network:443"
trust_height = 8959784
trust_hash = "3D8F12EA302AEDA66E80939F7FC785206692F8B6EE6F727F1655F1AFB6A873A5"
trust_period = "168h0m0s"

Start Gaia to begin state sync. It may take take some time for the node to acquire a snapshot, but the command and output should look similar to the following:

$ gaiad start --x-crisis-skip-assert-invariants

...

> INF Discovered new snapshot format=1 hash="0x000..." height=8967000 module=statesync

...

> INF Fetching snapshot chunk chunk=4 format=1 height=8967000 module=statesync total=45
> INF Applied snapshot chunk to ABCI app chunk=0 format=1 height=8967000 module=statesync total=45

Once state sync successfully completes, the node will begin to process blocks normally. If state sync fails and the node operator encounters the following error: State sync failed err="state sync aborted", either try restarting gaiad or running gaiad unsafe-reset-all (make sure to backup any configuration and history before doing this).

Quicksync

Quicksync.io offers several daily snapshots of the Cosmos Hub with varying levels of pruning (archive 1.4TB, default 540GB, and pruned 265GB). For downloads and installation instructions, visit the Cosmos Quicksync guide.

Snapshots

Saving and serving snapshots helps nodes rapidly join the network. Snapshots are now enabled by default effective 1/20/21.

While not advised, if a node operator needs to customize this feature, it can be configured in ~/.gaia/config/app.toml. The Cosmos Hub recommends setting this value to match pruning-keep-every in config.toml.

Note: It is highly recommended that node operators use the same value for snapshot-interval in order to aid snapshot discovery. Discovery is easier when more nodes are serving the same snapshots.

In app.toml

###############################################################################
### State Sync Configuration ###
###############################################################################

# State sync snapshots allow other nodes to rapidly join the network without replaying historical
# blocks, instead downloading and applying a snapshot of the application state at a given height.
[state-sync]

# snapshot-interval specifies the block interval at which local state sync snapshots are
# taken (0 to disable). Must be a multiple of pruning-keep-every.
snapshot-interval = 1000

# snapshot-keep-recent specifies the number of recent snapshots to keep and serve (0 to keep all).
snapshot-keep-recent = 10

Releases & Upgrades

See all Gaia Releases

The most up to date release of Gaia is above. For those that want to use state sync or quicksync to get their node up to speed, starting with the most recent version of Gaia is sufficient.

To sync an archive or full node from scratch, it is important to note that you must start with V4.2.1 and proceed through two different upgrades Delta at block height 6,910,000, Vega at block height 8,695,000, Theta at block height 10,085,397, Rho at block height 14099412 and Lambda at block height 14,470,501 and so on.

The process is summarized below but make sure to follow the manual upgrade instructions for each release:

Delta Instructions

Once V4 reaches the upgrade block height, expect the chain to halt and to see the following message:

ERR UPGRADE "Gravity-DEX" NEEDED at height: 6910000: v5.0.0-4760cf1f1266accec7a107f440d46d9724c6fd08

Make sure to save a backup of ~/.gaia in case rolling back is necessary.

Install Gaia V5.0.0 and restart the daemon.

Vega Instructions

Once V5 reaches the upgrade block height, the chain will halt and display the following message:

ERR UPGRADE "Vega" NEEDED at height: 8695000

Again, make sure to backup ~/.gaia

Install Gaia V6.0.0 and restart the daemon.

Theta Instructions

Once V6 reaches the upgrade block height, the chain will halt and display the following message:

ERR UPGRADE "Theta" NEEDED at height: 10085397

Again, make sure to backup ~/.gaia

Install Gaia V7.0.0 and restart the daemon.

Rho Instructions

Once V7 reaches the upgrade block height, the chain will halt and display the following message:

ERR UPGRADE "Rho" NEEDED at height: 14099412

Again, make sure to backup ~/.gaia

Install Gaia V8.0.0 and restart the daemon.

Lambda Instructions

Once V8 reaches the upgrade block height, the chain will halt and display the following message:

ERR UPGRADE "Lambda" NEEDED at height: 14470501

Again, make sure to backup ~/.gaia

Install Gaia V9.0.0 and restart the daemon.

Repeat the process for newer versions of the Gaia application at the stated block heights above.

Cosmovisor

Cosmovisor is a process manager developed to relieve node operators of having to manually intervene every time there is an upgrade. Cosmovisor monitors the governance module for upgrade proposals; it will take care of downloading the new binary, stopping the old one, switching to the new one, and restarting.

For more information on how to run a node via Cosmovisor, check out the docs.

Running via Background Process

To run the node in a background process with automatic restarts, it's recommended to use a service manager like systemd. To set this up run the following:

sudo tee /etc/systemd/system/<service name>.service > /dev/null <<EOF  
[Unit]
Description=Gaia Daemon
After=network-online.target

[Service]
User=$USER
ExecStart=$(which gaiad) start
Restart=always
RestartSec=3
LimitNOFILE=4096

[Install]
WantedBy=multi-user.target
EOF

If using Cosmovisor then make sure to add the following:

Environment="DAEMON_HOME=$HOME/.gaia"
Environment="DAEMON_NAME=gaiad"
Environment="DAEMON_ALLOW_DOWNLOAD_BINARIES=false"
Environment="DAEMON_RESTART_AFTER_UPGRADE=true"

After the LimitNOFILE line and replace $(which gaiad) with $(which cosmovisor).

Run the following to setup the daemon:

sudo -S systemctl daemon-reload
sudo -S systemctl enable <service name>

Then start the process and confirm that it's running.

sudo -S systemctl start <service name>

sudo service <service name> status

Exporting State

Gaia can dump the entire application state into a JSON file. This application state dump is useful for manual analysis and can also be used as the genesis file of a new network.

Note: The node can't be running while exporting state, otherwise the operator can expect a resource temporarily unavailable error.

Export state with:

gaiad export > [filename].json

It is also possible to export state from a particular height (at the end of processing the block of that height):

gaiad export --height [height] > [filename].json

If planning to start a new network from the exported state, export with the --for-zero-height flag:

gaiad export --height [height] --for-zero-height > [filename].json

Verify Mainnet

Help to prevent a catastrophe by running invariants on each block on your full node. In essence, by running invariants the node operator ensures that the state of mainnet is the correct expected state. One vital invariant check is that no atoms are being created or destroyed outside of expected protocol, however there are many other invariant checks each unique to their respective module. Because invariant checks are computationally expensive, they are not enabled by default. To run a node with these checks start your node with the --x-crisis-skip-assert-invariants flag:

gaiad start --x-crisis-skip-assert-invariants

If an invariant is broken on the node, it will panic and prompt the operator to send a transaction which will halt mainnet. For example the provided message may look like:

invariant broken:
loose token invariance:
pool.NotBondedTokens: 100
sum of account tokens: 101
CRITICAL please submit the following transaction:
gaiad tx crisis invariant-broken staking supply