cockroach start

On this page Carat arrow pointing down
Warning:
CockroachDB v21.1 is no longer supported as of November 18, 2022. For more details, refer to the Release Support Policy.

This page explains the cockroach start command, which you use to start a new multi-node cluster or add nodes to an existing cluster.

Tip:

If you need a simple single-node backend for app development, use cockroach start-single-node instead. For quick SQL testing, consider using cockroach demo to start a temporary, in-memory cluster with immediate access to an interactive SQL shell.

Note:

Node-level settings are defined by flags passed to the cockroach start command and cannot be changed without stopping and restarting the node. In contrast, some cluster-wide settings are defined via SQL statements and can be updated anytime after a cluster has been started. For more details, see Cluster Settings.

Synopsis

Start a node to be part of a new multi-node cluster:

icon/buttons/copy
$ cockroach start <flags, including --join>

Initialize a new multi-node cluster:

icon/buttons/copy
$ cockroach init <flags>

Add a node to an existing cluster:

icon/buttons/copy
$ cockroach start <flags, including --join>

View help:

icon/buttons/copy
$ cockroach start --help

Flags

The cockroach start command supports the following general-use, networking, security, and logging flags.

Many flags have useful defaults that can be overridden by specifying the flags explicitly. If you specify flags explicitly, however, be sure to do so each time the node is restarted, as they will not be remembered. The one exception is the --join flag, which is stored in a node's data directory. We still recommend specifying the --join flag every time, as this will allow nodes to rejoin the cluster even if their data directory was destroyed.

General

Flag Description
--attrs Arbitrary strings, separated by colons, specifying node capability, which might include specialized hardware or number of cores, for example:

--attrs=ram:64gb

These can be used to influence the location of data replicas. See Configure Replication Zones for full details.
--background Runs the node in the background. Control is returned to the shell only once the node is ready to accept requests, so this is recommended over appending & to the command. This flag is not available in Windows environments.

Note: --background is suitable for writing automated test suites or maintenance procedures that need a temporary server process running in the background. It is not intended to be used to start a long-running server, because it does not fully detach from the controlling terminal. Consider using a service manager or a tool like daemon(8) instead.
--cache The total size for caches, shared evenly if there are multiple storage devices. This can be a percentage (notated as a decimal or with %) or any bytes-based unit, for example:

--cache=.25
--cache=25%
--cache=1000000000 ----> 1000000000 bytes
--cache=1GB ----> 1000000000 bytes
--cache=1GiB ----> 1073741824 bytes

Note: If you use the % notation, you might need to escape the % sign, for instance, while configuring CockroachDB through systemd service files. For this reason, it's recommended to use the decimal notation instead.

Default: 128MiB

The default cache size is reasonable for local development clusters. For production deployments, this should be increased to 25% or higher. Increasing the cache size will generally improve the node's read performance. See Recommended Production Settings for more details.
--clock-device Enable CockroachDB to use a PTP hardware clock when querying the current time. The value is a string that specifies the clock device to use. For example: --clock-device=/dev/ptp0

Note: This is supported on Linux only and may be needed in cases where the host clock is unreliable or prone to large jumps (e.g., when using vMotion).
--cluster-name A string that specifies a cluster name. This is used together with --join to ensure that all newly created nodes join the intended cluster when you are running multiple clusters.

Note: If this is set, cockroach init, cockroach node decommission, cockroach node recommission, and the cockroach debug commands must specify either --cluster-name or --disable-cluster-name-verification in order to work.
--disable-cluster-name-verification On clusters for which a cluster name has been set, this flag paired with --cluster-name disables the cluster name check for the command. This is necessary on existing clusters, when setting a cluster name or changing the cluster name: Perform a rolling restart of all nodes and include both the new --cluster-name value and --disable-cluster-name-verification, then a second rolling restart with --cluster-name and without --disable-cluster-name-verification.
--external-io-dir The path of the external IO directory with which the local file access paths are prefixed while performing backup and restore operations using local node directories or NFS drives. If set to disabled, backups and restores using local node directories and NFS drives, as well as cockroach nodelocal upload, are disabled.

Default: extern subdirectory of the first configured store.

To set the --external-io-dir flag to the locations you want to use without needing to restart nodes, create symlinks to the desired locations from within the extern directory.
--listening-url-file The file to which the node's SQL connection URL will be written as soon as the node is ready to accept connections, in addition to being printed to the standard output. When --background is used, this happens before the process detaches from the terminal.

This is particularly helpful in identifying the node's port when an unused port is assigned automatically (--port=0).
--locality Arbitrary key-value pairs that describe the location of the node. Locality might include country, region, availability zone, etc. A region tier must be included in order to enable multi-region capabilities. For more details, see Locality below.
--max-disk-temp-storage The maximum on-disk storage capacity available to store temporary data for SQL queries that exceed the memory budget (see --max-sql-memory). This ensures that JOINs, sorts, and other memory-intensive SQL operations are able to spill intermediate results to disk. This can be a percentage (notated as a decimal or with %) or any bytes-based unit (e.g., .25, 25%, 500GB, 1TB, 1TiB).

Note: If you use the % notation, you might need to escape the % sign, for instance, while configuring CockroachDB through systemd service files. For this reason, it's recommended to use the decimal notation instead. Also, if expressed as a percentage, this value is interpreted relative to the size of the first store. However, the temporary space usage is never counted towards any store usage; therefore, when setting this value, it's important to ensure that the size of this temporary storage plus the size of the first store doesn't exceed the capacity of the storage device.

The temporary files are located in the path specified by the --temp-dir flag, or in the subdirectory of the first store (see --store) by default.

Default: 32GiB
--max-offset The maximum allowed clock offset for the cluster. If observed clock offsets exceed this limit, servers will crash to minimize the likelihood of reading inconsistent data. Increasing this value will increase the time to recovery of failures as well as the frequency of uncertainty-based read restarts.

Note that this value must be the same on all nodes in the cluster and cannot be changed with a rolling upgrade. In order to change it, first stop every node in the cluster. Then once the entire cluster is offline, restart each node with the new value.

Default: 500ms
--max-sql-memory The maximum in-memory storage capacity available to store temporary data for SQL queries, including prepared queries and intermediate data rows during query execution. This can be a percentage (notated as a decimal or with %) or any bytes-based unit, for example:

--max-sql-memory=.25
--max-sql-memory=25%
--max-sql-memory=10000000000 ----> 1000000000 bytes
--max-sql-memory=1GB ----> 1000000000 bytes
--max-sql-memory=1GiB ----> 1073741824 bytes

The temporary files are located in the path specified by the --temp-dir flag, or in the subdirectory of the first store (see --store) by default.

Note: If you use the % notation, you might need to escape the % sign, for instance, while configuring CockroachDB through systemd service files. For this reason, it's recommended to use the decimal notation instead.

Default: 25%

The default SQL memory size is suitable for production deployments but can be raised to increase the number of simultaneous client connections the node allows as well as the node's capacity for in-memory processing of rows when using ORDER BY, GROUP BY, DISTINCT, joins, and window functions. For local development clusters with memory-intensive workloads, reduce this value to, for example, 128MiB to prevent out of memory errors.
--pid-file The file to which the node's process ID will be written as soon as the node is ready to accept connections. When --background is used, this happens before the process detaches from the terminal. When this flag is not set, the process ID is not written to file.
--store
-s
The file path to a storage device and, optionally, store attributes and maximum size. When using multiple storage devices for a node, this flag must be specified separately for each device, for example:

--store=/mnt/ssd01 --store=/mnt/ssd02

For more details, see Store below.
--spatial-libs The location on disk where CockroachDB looks for spatial libraries.

Defaults:
  • /usr/local/lib/cockroach
  • A lib subdirectory of the CockroachDB binary's current directory.

--temp-dir The path of the node's temporary store directory. On node start up, the location for the temporary files is printed to the standard output.

Default: Subdirectory of the first store

Networking

Flag Description
--experimental-dns-srv When this flag is included, the node will first attempt to fetch SRV records from DNS for every name specified with --join. If a valid SRV record is found, that information is used instead of regular DNS A/AAAA lookups. This feature is experimental and may be removed or modified in a later version.
--listen-addr The IP address/hostname and port to listen on for connections from other nodes and clients. For IPv6, use the notation [...], e.g., [::1] or [fe80::f6f2:::].

This flag's effect depends on how it is used in combination with --advertise-addr. For example, the node will also advertise itself to other nodes using this value if --advertise-addr is not specified. For more details, see Networking.

Default: Listen on all IP addresses on port 26257; if --advertise-addr is not specified, also advertise the node's canonical hostname to other nodes
--advertise-addr The IP address/hostname and port to tell other nodes to use. If using a hostname, it must be resolvable from all nodes. If using an IP address, it must be routable from all nodes; for IPv6, use the notation [...], e.g., [::1] or [fe80::f6f2:::].

This flag's effect depends on how it is used in combination with --listen-addr. For example, if the port number is different than the one used in --listen-addr, port forwarding is required. For more details, see Networking.

Default: The value of --listen-addr; if --listen-addr is not specified, advertises the node's canonical hostname and port 26257
--http-addr The IP address/hostname and port to listen on for DB Console HTTP requests. For IPv6, use the notation [...], e.g., [::1]:8080 or [fe80::f6f2:::]:8080.

Default: Listen on the address part of --listen-addr on port 8080
--locality-advertise-addr The IP address/hostname and port to tell other nodes in specific localities to use. This flag is useful when running a cluster across multiple networks, where nodes in a given network have access to a private or local interface while nodes outside the network do not. In this case, you can use --locality-advertise-addr to tell nodes within the same network to prefer the private or local address to improve performance and use --advertise-addr to tell nodes outside the network to use another address that is reachable from them.

This flag relies on nodes being started with the --locality flag and uses the locality@address notation, for example:

--locality-advertise-addr=region=us-west@10.0.0.0:26257

See the example below for more details.
--sql-addr The IP address/hostname and port to listen on for SQL connections from clients. For IPv6, use the notation [...], e.g., [::1] or [fe80::f6f2:::].

This flag's effect depends on how it is used in combination with --advertise-sql-addr. For example, the node will also advertise itself to clients using this value if --advertise-sql-addr is not specified.

Default: The value of --listen-addr; if --listen-addr is not specified, advertises the node's canonical hostname and port 26257

For an example, see Start a cluster with separate RPC and SQL networks
--advertise-sql-addr The IP address/hostname and port to tell clients to use. If using a hostname, it must be resolvable from all nodes. If using an IP address, it must be routable from all nodes; for IPv6, use the notation [...], e.g., [::1] or [fe80::f6f2:::].

This flag's effect depends on how it is used in combination with --sql-addr. For example, if the port number is different than the one used in --sql-addr, port forwarding is required.

Default: The value of --sql-addr; if --sql-addr is not specified, advertises the value of --listen-addr
--join
-j
The host addresses that connect nodes to the cluster and distribute the rest of the node addresses. These can be IP addresses or DNS aliases of nodes.

When starting a cluster in a single region, specify the addresses of 3-5 initial nodes. When starting a cluster in multiple regions, specify more than 1 address per region, and select nodes that are spread across failure domains. Then run the cockroach init command against any of these nodes to complete cluster startup. See the example below for more details.

Use the same --join list for all nodes to ensure that the cluster can stabilize. Do not list every node in the cluster, because this increases the time for a new cluster to stabilize. Note that these are best practices; it is not required to restart an existing node to update its --join flag.

cockroach start must be run with the --join flag. To start a single-node cluster, use cockroach start-single-node instead.
--socket-dir The directory path on which to listen for Unix domain socket connections from clients installed on the same Unix-based machine. For an example, see Connect to a cluster listening for Unix domain socket connections.
--advertise-host Deprecated. Use --advertise-addr instead.
--host Deprecated. Use --listen-addr instead.
--port
-p
Deprecated. Specify port in --advertise-addr and/or --listen-addr instead.
--http-host Deprecated. Use --http-addr instead.
--http-port Deprecated. Specify port in --http-addr instead.

Security

Flag Description
--certs-dir The path to the certificate directory. The directory must contain valid certificates if running in secure mode.

Default: ${HOME}/.cockroach-certs/
--insecure Note: The --insecure flag is intended for non-production testing only.

Run in insecure mode, skipping all TLS encryption and authentication. If this flag is not set, the --certs-dir flag must point to valid certificates.

Note the following risks: An insecure cluster is open to any client that can access any node's IP addresses; client connections must also be made insecurely; any user, even root, can log in without providing a password; any user, connecting as root, can read or write any data in your cluster; there is no network encryption or authentication, and thus no confidentiality.

Default: false
--accept-sql-without-tls This experimental flag allows you to connect to the cluster using a SQL user's password without validating the client's certificate. When connecting using the built-in SQL client, use the --insecure flag with the cockroach sql command.
--cert-principal-map A comma-separated list of cert-principal:db-principal mappings used to map the certificate principals to IP addresses, DNS names, and SQL users. This allows the use of certificates generated by Certificate Authorities that place restrictions on the contents of the commonName field. For usage information, see Create Security Certificates using Openssl.
--enterprise-encryption This optional flag specifies the encryption options for one of the stores on the node. If multiple stores exist, the flag must be specified for each store.

This flag takes a number of options. For a complete list of options, and usage instructions, see Encryption at Rest.

Note that this is an Enterprise feature.
--external-io-disable-http This optional flag disables external HTTP(S) access (as well as custom HTTP(S) endpoints) when performing bulk operations (e.g, BACKUP, IMPORT, etc.). This can be used in environments where you cannot run a full proxy server.

If you want to run a proxy server, you can start CockroachDB while specifying the HTTP(S)_PROXY environment variable.
--external-io-disable-implicit-credentials This optional flag disables the use of implicit credentials when accessing external cloud storage services for bulk operations (e.g, BACKUP, IMPORT, etc.).

Locality

The --locality flag accepts arbitrary key-value pairs that describe the location of the node. Locality might include region, country, availability zone, etc. The key-value pairs should be ordered into locality tiers from most inclusive to least inclusive (e.g., region before availability zone as in region=eu,zone=paris), and the keys and order of key-value pairs must be the same on all nodes. It's typically better to include more pairs than fewer.

  • Specifying a region with a region tier is required in order to enable CockroachDB's multi-region capabilities.

  • CockroachDB spreads the replicas of each piece of data across as diverse a set of localities as possible, with the order determining the priority. Locality can also be used to influence the location of data replicas in various ways using replication zones.

  • When there is high latency between nodes (e.g., cross-availability zone deployments), CockroachDB uses locality to move range leases closer to the current workload, reducing network round trips and improving read performance, also known as "follow-the-workload". In a deployment across more than 3 availability zones, however, to ensure that all data benefits from "follow-the-workload", you must increase your replication factor to match the total number of availability zones.

  • Locality is also a prerequisite for using the table partitioning and Node Map Enterprise features.

Example

icon/buttons/copy
# Locality flag for nodes in US East availability zone:
--locality=region=us,zone=us-east

# Locality flag for nodes in US Central availability zone:
--locality=region=us,zone=us-central

# Locality flag for nodes in US West availability zone:
--locality=region=us,zone=us-west

Storage

Storage engine

The --storage-engine flag is used to choose the storage engine used by the node. Note that this setting applies to all stores on the node, including the temp store.

New in v21.1: As of v21.1, CockroachDB always uses the Pebble storage engine. As such, pebble is the default and only option for the --storage-engine flag.

Warning:

If you specified --storage-engine=rocksdb when starting nodes with v20.2, be sure to change that to --storage-engine=pebble or remove the flag entirely before upgrading to v21.1. Pebble is intended to be bi-directionally compatible with the RocksDB on-disk format, so the switch will be seamless during the upgrade process.

Store

The --store flag supports the following fields. Note that commas are used to separate fields, and so are forbidden in all field values.

Note:

In-memory storage is not suitable for production deployments at this time.

Field Description
type For in-memory storage, set this field to mem; otherwise, leave this field out. The path field must not be set when type=mem.
path The file path to the storage device. When not setting attr or size, the path field label can be left out:

--store=/mnt/ssd01

When either of those fields are set, however, the path field label must be used:

--store=path=/mnt/ssd01,size=20GB

Default: cockroach-data
attrs Arbitrary strings, separated by colons, specifying disk type or capability. These can be used to influence the location of data replicas. See Configure Replication Zones for full details.

In most cases, node-level --locality or --attrs are preferable to store-level attributes, but this field can be used to match capabilities for storage of individual databases or tables. For example, an OLTP database would probably want to allocate space for its tables only on solid state devices, whereas append-only time series might prefer cheaper spinning drives. Typical attributes include whether the store is flash (ssd) or spinny disk (hdd), as well as speeds and other specs, for example:

--store=path=/mnt/hda1,attrs=hdd:7200rpm
size The maximum size allocated to the node. When this size is reached, CockroachDB attempts to rebalance data to other nodes with available capacity. When there's no capacity elsewhere, this limit will be exceeded. Also, data may be written to the node faster than the cluster can rebalance it away; in this case, as long as capacity is available elsewhere, CockroachDB will gradually rebalance data down to the store limit.

The size can be specified either in a bytes-based unit or as a percentage of hard drive space (notated as a decimal or with %), for example:

--store=path=/mnt/ssd01,size=10000000000 ----> 10000000000 bytes
--store=path=/mnt/ssd01,size=20GB ----> 20000000000 bytes
--store=path=/mnt/ssd01,size=20GiB ----> 21474836480 bytes
--store=path=/mnt/ssd01,size=0.02TiB ----> 21474836480 bytes
--store=path=/mnt/ssd01,size=20% ----> 20% of available space
--store=path=/mnt/ssd01,size=0.2 ----> 20% of available space
--store=path=/mnt/ssd01,size=.2 ----> 20% of available space

Default: 100%

For an in-memory store, the size field is required and must be set to the true maximum bytes or percentage of available memory, for example:

--store=type=mem,size=20GB
--store=type=mem,size=90%

Note: If you use the % notation, you might need to escape the % sign, for instance, while configuring CockroachDB through systemd service files. For this reason, it's recommended to use the decimal notation instead.

Logging

By default, cockroach start writes all messages to log files, and prints nothing to stderr. This includes events with INFO severity and higher. However, you can customize the logging behavior of this command by using the --log flag:

Flag Description
--log New in v21.1: Configure logging parameters by specifying a YAML payload. For details, see Configure logs. If a YAML configuration is not specified, the default configuration is used.

--log-config-file can also be used.

Note: The deprecated logging flags below cannot be combined with --log, and can be defined instead in the YAML payload.
--log-config-file New in v21.1: Configure logging parameters by specifying a path to a YAML file. For details, see Configure logs. If a YAML configuration is not specified, the default configuration is used.

--log can also be used.

Note: The deprecated logging flags below cannot be combined with --log-config-file, and can be defined instead in the YAML payload.
--log-dir Deprecated. To enable logging to files and write logs to the specified directory, use --log and set dir in the YAML configuration.

Setting --log-dir to a blank directory (--log-dir=) disables logging to files. Do not use --log-dir=""; this creates a new directory named "" and stores log files in that directory.
--log-group-max-size Deprecated. This is now configured with --log or --log-config-file and a YAML payload. After the logging group (i.e., cockroach, cockroach-sql-audit, cockroach-auth, cockroach-sql-exec, cockroach-pebble) reaches the specified size, delete the oldest log file. The flag's argument takes standard file sizes, such as --log-group-max-size=1GiB.

Default: 100MiB
--log-file-max-size Deprecated. This is now configured with --log or --log-config-file and a YAML payload. After logs reach the specified size, begin writing logs to a new file. The flag's argument takes standard file sizes, such as --log-file-max-size=2MiB.

Default: 10MiB
--log-file-verbosity Deprecated. This is now configured with --log or --log-config-file and a YAML payload. Only writes messages to log files if they are at or above the specified severity level, such as --log-file-verbosity=WARNING. Requires logging to files.

Default: INFO
--logtostderr Deprecated. This is now configured with --log or --log-config-file and a YAML payload. Enable logging to stderr for messages at or above the specified severity level, such as --logtostderr=ERROR

If you use this flag without specifying the severity level (e.g., cockroach start --logtostderr), it prints messages of all severities to stderr.

Setting --logtostderr=NONE disables logging to stderr.
--no-color Do not colorize stderr. Possible values: true or false.

When set to false, messages logged to stderr are colorized based on severity level.

Default: false
--sql-audit-dir Deprecated. This is now configured with --log or --log-config-file and a YAML payload. If non-empty, output the SENSITIVE_ACCESS logging channel to this directory.

Note that enabling SENSITIVE_ACCESS logs can negatively impact performance. As a result, we recommend using the SENSITIVE_ACCESS channel for security purposes only. For more information, see Logging use cases.

Defaults

See the default logging configuration.

Standard output

When you run cockroach start, some helpful details are printed to the standard output:

CockroachDB node starting at 
build:               CCL v21.1.21 @ 2022-09-15 00:00:00 (go1.12.6)
webui:               http://localhost:8080
sql:                 postgresql://root@localhost:26257?sslmode=disable
RPC client flags:    cockroach <client cmd> --host=localhost:26257 --insecure
logs:                /Users/<username>/node1/logs
temp dir:            /Users/<username>/node1/cockroach-temp242232154
external I/O path:   /Users/<username>/node1/extern
store[0]:            path=/Users/<username>/node1
status:              initialized new cluster
clusterID:           8a681a16-9623-4fc1-a537-77e9255daafd
nodeID:              1
Tip:

These details are also written to the INFO log in the /logs directory. You can retrieve them with a command like grep 'node starting' node1/logs/cockroach.log -A 11.

Field Description
build The version of CockroachDB you are running.
webui The URL for accessing the DB Console.
sql The connection URL for your client.
RPC client flags The flags to use when connecting to the node via cockroach client commands.
logs The directory containing debug log data.
temp dir The temporary store directory of the node.
external I/O path The external IO directory with which the local file access paths are prefixed while performing backup and restore operations using local node directories or NFS drives.
attrs If node-level attributes were specified in the --attrs flag, they are listed in this field. These details are potentially useful for configuring replication zones.
locality If values describing the locality of the node were specified in the --locality field, they are listed in this field. These details are potentially useful for configuring replication zones.
store[n] The directory containing store data, where [n] is the index of the store, e.g., store[0] for the first store, store[1] for the second store.

If store-level attributes were specified in the attrs field of the --store flag, they are listed in this field as well. These details are potentially useful for configuring replication zones.
status Whether the node is the first in the cluster (initialized new cluster), joined an existing cluster for the first time (initialized new node, joined pre-existing cluster), or rejoined an existing cluster (restarted pre-existing node).
clusterID The ID of the cluster.

When trying to join a node to an existing cluster, if this ID is different than the ID of the existing cluster, the node has started a new cluster. This may be due to conflicting information in the node's data directory. For additional guidance, see the troubleshooting docs.
nodeID The ID of the node.

Examples

Start a multi-node cluster

To start a multi-node cluster, run the cockroach start command for each node, setting the --join flag to the addresses of the initial nodes.

For a cluster in a single region, set 3-5 --join addresses. Each starting node will attempt to contact one of the join hosts. In case a join host cannot be reached, the node will try another address on the list until it can join the gossip network.

Note:

When starting a multi-region cluster, set more than one --join address per region, and select nodes that are spread across failure domains. This ensures high availability.

Tip:

Before starting the cluster, use cockroach cert to generate node and client certificates for a secure cluster connection.

icon/buttons/copy
$ cockroach start \
--certs-dir=certs \
--advertise-addr=<node1 address> \
--join=<node1 address>,<node2 address>,<node3 address> \
--cache=.25 \
--max-sql-memory=.25
icon/buttons/copy
$ cockroach start \
--certs-dir=certs \
--advertise-addr=<node2 address> \
--join=<node1 address>,<node2 address>,<node3 address> \
--cache=.25 \
--max-sql-memory=.25
icon/buttons/copy
$ cockroach start \
--certs-dir=certs \
--advertise-addr=<node3 address> \
--join=<node1 address>,<node2 address>,<node3 address> \
--cache=.25 \
--max-sql-memory=.25

icon/buttons/copy

$ cockroach start \
--insecure \
--advertise-addr=<node1 address> \
--join=<node1 address>,<node2 address>,<node3 address> \
--cache=.25 \
--max-sql-memory=.25
icon/buttons/copy
$ cockroach start \
--insecure \
--advertise-addr=<node2 address> \
--join=<node1 address>,<node2 address>,<node3 address> \
--cache=.25 \
--max-sql-memory=.25
icon/buttons/copy
$ cockroach start \
--insecure \
--advertise-addr=<node3 address> \
--join=<node1 address>,<node2 address>,<node3 address> \
--cache=.25 \
--max-sql-memory=.25

Then run the cockroach init command against any node to perform a one-time cluster initialization:

icon/buttons/copy

$ cockroach init \
--certs-dir=certs \
--host=<address of any node>

icon/buttons/copy

$ cockroach init \
--insecure \
--host=<address of any node>

Start a multi-node cluster across private networks

Scenario:

  • You have a cluster that spans GCE and AWS.
  • The nodes on each cloud can reach each other on private addresses, but the private addresses aren't reachable from the other cloud.

Approach:

  1. Start each node on GCE with --locality set to describe its location, --locality-advertise-addr set to advertise its private address to other nodes in on GCE, --advertise-addr set to advertise its public address to nodes on AWS, and --join set to the public addresses of 3-5 of the initial nodes:

    icon/buttons/copy
    $ cockroach start \
    --certs-dir=certs \
    --locality=cloud=gce \
    --locality-advertise-addr=cloud=gce@<private address of node> \
    --advertise-addr=<public address of node> \
    --join=<public address of node1>,<public address of node2>,<public address of node3> \
    --cache=.25 \
    --max-sql-memory=.25
    
  2. Start each node on AWS with --locality set to describe its location, --locality-advertise-addr set to advertise its private address to other nodes on AWS, --advertise-addr set to advertise its public address to nodes on GCE, and --join set to the public addresses of 3-5 of the initial nodes:

    icon/buttons/copy
    $ cockroach start \
    --certs-dir=certs \
    --locality=cloud=aws \
    --locality-advertise-addr=cloud=aws@<private address of node> \
    --advertise-addr=<public address of node> \
    --join=<public address of node1>,<public address of node2>,<public address of node3> \
    --cache=.25 \
    --max-sql-memory=.25
    
  3. Run the cockroach init command against any node to perform a one-time cluster initialization:

    icon/buttons/copy
    $ cockroach init \
    --certs-dir=certs \
    --host=<address of any node>
    

Add a node to a cluster

To add a node to an existing cluster, run the cockroach start command, setting the --join flag to the same addresses you used when starting the cluster:

icon/buttons/copy

$ cockroach start \
--certs-dir=certs \
--advertise-addr=<node4 address> \
--join=<node1 address>,<node2 address>,<node3 address> \
--cache=.25 \
--max-sql-memory=.25

icon/buttons/copy

$ cockroach start \
--insecure \
--advertise-addr=<node4 address> \
--join=<node1 address>,<node2 address>,<node3 address> \
--cache=.25 \
--max-sql-memory=.25

Create a table with node locality information

Start a three-node cluster with locality information specified in the cockroach start commands:

icon/buttons/copy
$ cockroach start --insecure --port=26257 --http-port=26258 --store=cockroach-data/1 --cache=256MiB --locality=region=eu-west-1,cloud=aws,zone=eu-west-1a
icon/buttons/copy
$ cockroach start --insecure --port=26259 --http-port=26260 --store=cockroach-data/2 --cache=256MiB --join=localhost:26257 --locality=region=eu-west-1,cloud=aws,zone=eu-west-1b
icon/buttons/copy
$ cockroach start --insecure --port=26261 --http-port=26262 --store=cockroach-data/3 --cache=256MiB --join=localhost:26257 --locality=region=eu-west-1,cloud=aws,zone=eu-west-1c

You can use the crdb_internal.locality_value built-in function to return the current node's locality information from inside a SQL shell. The example below uses the output of crdb_internal.locality_value('zone') as the DEFAULT value to use for the zone column of new rows. Other available locality keys for the running three-node cluster include region and cloud.

icon/buttons/copy
$ cockroach sql --insecure
icon/buttons/copy
> CREATE TABLE charges (
  zone STRING NOT NULL DEFAULT crdb_internal.locality_value('zone'),
  id INT PRIMARY KEY NOT NULL
);
icon/buttons/copy
> INSERT INTO charges (id) VALUES (1);
icon/buttons/copy
> SELECT * FROM charges WHERE id = 1;
     zone    | id
+------------+----+
  eu-west-1a |  1
(1 row)

The zone column has the zone of the node on which the row was created.

In a separate terminal window, open a SQL shell to a different node on the cluster:

icon/buttons/copy
$ cockroach sql --insecure --port 26259
icon/buttons/copy
> INSERT INTO charges (id) VALUES (2);
icon/buttons/copy
> SELECT * FROM charges WHERE id = 2;
     zone    | id
+------------+----+
  eu-west-1b |  2
(1 row)

In a separate terminal window, open a SQL shell to the third node:

icon/buttons/copy
$ cockroach sql --insecure --port 26261
icon/buttons/copy
> INSERT INTO charges (id) VALUES (3);
icon/buttons/copy
> SELECT * FROM charges WHERE id = 3;
     zone    | id
+------------+----+
  eu-west-1c |  3
(1 row)

Start a cluster with separate RPC and SQL networks

Separating the network addresses used for intra-cluster RPC traffic and application SQL connections can provide an additional level of protection against security issues as a form of defense in depth. This separation is accomplished with a combination of the --sql-addr flag and firewall rules or other network-level access control (which must be maintained outside of CockroachDB).

For example, suppose you want to use port 26257 for SQL connections and 26258 for intra-cluster traffic. Set up firewall rules so that the CockroachDB nodes can reach each other on port 26258, but other machines cannot. Start the CockroachDB processes as follows:

icon/buttons/copy
$ cockroach start --sql-addr=:26257 --listen-addr=:26258 --join=node1:26258,node2:26258,node3:26258 --certs-dir=~/cockroach-certs

Note the use of port 26258 (the value for listen-addr, not sql-addr) in the --join flag. Also, if your environment requires the use of the --advertise-addr flag, you should probably also use the --advertise-sql-addr flag when using a separate SQL address.

Clusters using this configuration with client certificate authentication may also wish to use split client CA certificates.

See also


Yes No
On this page

Yes No