Get started with InfluxDB 3 Core

Get started with InfluxDB 3 Core

InfluxDB is a database built to collect, process, transform, and store event and time series data, and is ideal for use cases that require real-time ingest and fast query response times to build user interfaces, monitoring, and automation solutions.

Common use cases include:

• Monitoring sensor data
• Server monitoring
• Application performance monitoring
• Network monitoring
• Financial market and trading analytics
• Behavioral analytics

InfluxDB is optimized for scenarios where near real-time data monitoring is essential and queries need to return quickly to support user experiences such as dashboards and interactive user interfaces.

InfluxDB 3 Core is the InfluxDB 3 open source release. Core’s feature highlights include:

• Diskless architecture with object storage support (or local disk with no dependencies)
• Fast query response times (under 10ms for last-value queries, or 30ms for distinct metadata)
• Embedded Python VM for plugins and triggers
• Parquet file persistence
• Compatibility with InfluxDB 1.x and 2.x write APIs

The Enterprise version adds the following features to Core:

• Historical query capability and single series indexing
• High availability
• Read replicas
• Enhanced security (coming soon)
• Row-level delete support (coming soon)
• Integrated admin UI (coming soon)

For more information, see how to get started with Enterprise.

What’s in this guide

This guide covers InfluxDB 3 Core (the open source release), including the following topics:

• Install and startup
• Data Model
• Write data to the database
• Query the database
• Last values cache
• Distinct values cache
• Python plugins and the processing engine

Install and startup

InfluxDB 3 Core runs on Linux, macOS, and Windows.

curl -O https://www.influxdata.com/d/install_influxdb3.sh \
&& sh install_influxdb3.sh

docker pull quay.io/influxdb/influxdb3-core:latest

Build artifacts and images update with every merge into the InfluxDB 3 Core main branch.

Verify the install

After you have installed InfluxDB 3 Core, enter the following command to verify that it completed successfully:

influxdb3 --version

If your system doesn’t locate influxdb3, then source the configuration file (for example, .bashrc, .zshrc) for your shell–for example:

source ~/.zshrc

Start InfluxDB

To start your InfluxDB instance, use the influxdb3 serve command and provide the following:

• --object-store: Specifies the type of Object store to use. InfluxDB supports the following: local file system (file), memory, S3 (and compatible services like Ceph or Minio) (s3), Google Cloud Storage (google), and Azure Blob Storage (azure).
• --node-id: A string identifier that determines the server’s storage path within the configured storage location, and, in a multi-node setup, is used to reference the node.

The following examples show how to start InfluxDB 3 with different object store configurations:

# Memory object store
# Stores data in RAM; doesn't persist data
influxdb3 serve --node-id=local01 --object-store=memory

# Filesystem object store
# Provide the filesystem directory
influxdb3 serve \
  --node-id=local01 \
  --object-store=file \
  --data-dir ~/.influxdb3

To run the Docker image and persist data to the filesystem, mount a volume for the Object store-for example, pass the following options:

• -v /path/on/host:/path/in/container: Mounts a directory from your filesystem to the container
• --object-store file --data-dir /path/in/container: Uses the mount for server storage

# Filesystem object store with Docker 
# Create a mount
# Provide the mount path

docker run -it \
 -v /path/on/host:/path/in/container \
 quay.io/influxdb/influxdb3-core:latest serve \
 --node-id my_host \
 --object-store file \
 --data-dir /path/in/container

# S3 object store (default is the us-east-1 region)
# Specify the Object store type and associated options

```bash
influxdb3 serve \
  --node-id=local01 \
  --object-store=s3 \
  --bucket=BUCKET \
  --aws-access-key=AWS_ACCESS_KEY \
  --aws-secret-access-key=AWS_SECRET_ACCESS_KEY

# Minio or other open source object store
# (using the AWS S3 API with additional parameters)
# Specify the object store type and associated options

```bash
influxdb3 serve --node-id=local01 --object-store=s3 --bucket=BUCKET \
  --aws-access-key=AWS_ACCESS_KEY \
  --aws-secret-access-key=AWS_SECRET_ACCESS_KEY \
  --aws-endpoint=ENDPOINT \
  --aws-allow-http

For more information about server options, run influxdb3 serve --help.

Data model

The database server contains logical databases, which have tables, which have columns. Compared to previous versions of InfluxDB you can think of a database as a bucket in v2 or as a db/retention_policy in v1. A table is equivalent to a measurement, which has columns that can be of type tag (a string dictionary), int64, float64, uint64, bool, or string and finally every table has a time column that is a nanosecond precision timestamp.

In InfluxDB 3, every table has a primary key–the ordered set of tags and the time–for its data. This is the sort order used for all Parquet files that get created. When you create a table, either through an explicit call or by writing data into a table for the first time, it sets the primary key to the tags in the order they arrived. This is immutable. Although InfluxDB is still a schema-on-write database, the tag column definitions for a table are immutable.

Tags should hold unique identifying information like sensor_id, or building_id or trace_id. All other data should be kept in fields. You will be able to add fast last N value and distinct value lookups later for any column, whether it is a field or a tag.

Write data

InfluxDB is a schema-on-write database. You can start writing data and InfluxDB creates the logical database, tables, and their schemas on the fly. After a schema is created, InfluxDB validates future write requests against it before accepting the data. Subsequent requests can add new fields on-the-fly, but can’t add new tags.

InfluxDB 3 Core provides three write API endpoints that respond to HTTP POST requests:

/api/v3/write_lp endpoint

InfluxDB 3 Core adds the /api/v3/write_lp endpoint.

POST /api/v3/write_lp?db=mydb&precision=nanosecond&accept_partial=true

This endpoint accepts the same line protocol syntax as previous versions, and supports the ?accept_partial=<BOOLEAN> parameter, which lets you accept or reject partial writes (default is true).

/api/v2/write InfluxDB v2 compatibility endpoint

Provides backwards compatibility with clients that can write data to InfluxDB OSS v2.x and Cloud 2 (TSM).

POST /api/v2/write?bucket=mydb&precision=ns

/write InfluxDB v1 compatibility endpoint

Provides backwards compatibility for clients that can write data to InfluxDB v1.x

POST /write?db=mydb&precision=ns

Keep in mind that these compatibility APIs differ from the v1 and v2 APIs in previous versions in the following ways:

• Tags in a table (measurement) are immutable
• A tag and a field can’t have the same name within a table.

Write line protocol

The following code block is an example of time series data in line protocol syntax:

• cpu: the table name.
• host, region, applications: the tags. A tag set is an ordered, comma-separated list of key/value pairs where the values are strings.
• val, usage_percent, status: the fields. A field set is a comma-separated list of key/value pairs.
• timestamp: If you don’t specify a timestamp, InfluxData uses the time when data is written. The default precision is a nanosecond epoch. To specify a different precision, pass the precision query parameter.

cpu,host=Alpha,region=us-west,application=webserver val=1i,usage_percent=20.5,status="OK"
cpu,host=Bravo,region=us-east,application=database val=2i,usage_percent=55.2,status="OK"
cpu,host=Charlie,region=us-west,application=cache val=3i,usage_percent=65.4,status="OK"
cpu,host=Bravo,region=us-east,application=database val=4i,usage_percent=70.1,status="Warn"
cpu,host=Bravo,region=us-central,application=database val=5i,usage_percent=80.5,status="OK"
cpu,host=Alpha,region=us-west,application=webserver val=6i,usage_percent=25.3,status="Warn"

Example: write data using the influxdb3 CLI

If you save the preceding line protocol to a file (for example, server_data), then you can use the influxdb3 CLI to write the data–for example:

influxdb3 write --database=mydb --file=server_data

Example: write data using the /api/v3 HTTP API

The following examples show how to write data using curl and the /api/3/write_lp HTTP endpoint. To show the difference between accepting and rejecting partial writes, line 2 in the example contains a string value for a float field (temp=hi).

Partial write of line protocol occurred

With accept_partial=true:

* upload completely sent off: 59 bytes
< HTTP/1.1 400 Bad Request
< transfer-encoding: chunked
< date: Wed, 15 Jan 2025 19:35:36 GMT
< 
* Connection #0 to host localhost left intact
{
  "error": "partial write of line protocol occurred",
  "data": [
    {
      "original_line": "dquote> home,room=Sunroom temp=hi",
      "line_number": 2,
      "error_message": "No fields were provided"
    }
  ]
}

Line 1 is written and queryable. The response is an HTTP error (400) status, and the response body contains the error message partial write of line protocol occurred with details about the problem line.

Parsing failed for write_lp endpoint

With accept_partial=false:

curl -v "http://localhost:8181/api/v3/write_lp?db=sensors&precision=auto&accept_partial=false" \
  --data-raw "home,room=Sunroom temp=96
home,room=Sunroom temp=hi"

The response is the following:

< HTTP/1.1 400 Bad Request
< transfer-encoding: chunked
< date: Wed, 15 Jan 2025 19:28:27 GMT
< 
* Connection #0 to host localhost left intact
{
  "error": "parsing failed for write_lp endpoint",
  "data": {
    "original_line": "home,room=Sunroom temp=hi",
    "line_number": 2,
    "error_message": "No fields were provided"
  }
}

InfluxDB rejects all points in the batch. The response is an HTTP error (400) status, and the response body contains parsing failed for write_lp endpoint and details about the problem line.

Data flow

The figure below shows how written data flows through the database.

1. Incoming writes: The system validates incoming data and stores it in the write buffer (in memory). If no_sync=true, the server sends a response to acknowledge the write.
2. WAL flush: Every second (default), the system flushes the write buffer to the Write-Ahead Log (WAL) for persistence in the Object store. If no_sync=false (default), the server sends a response to acknowledge the write.
3. Query availability: After WAL persistence completes, data moves to the queryable buffer where it becomes available for queries. By default, the server keeps up to 900 WAL files (15 minutes of data) buffered.
4. Long-term storage in Parquet: Every ten minutes (default), the system persists the oldest data from the queryable buffer to the Object store in Parquet format. InfluxDB keeps the remaining data (the most recent 5 minutes) in memory.
5. In-memory cache: InfluxDB puts Parquet files into an in-memory cache so that queries against the most recently persisted data don’t have to go to object storage.

Write responses

By default, InfluxDB acknowledges writes after flushing the WAL file to the Object store (occurring every second). For high write throughput, you can send multiple concurrent write requests.

Use no_sync for immediate write responses

To reduce the latency of writes, use the no_sync write option, which acknowledges writes before WAL persistence completes. When no_sync=true, InfluxDB validates the data, writes the data to the WAL, and then immediately responds to the client, without waiting for persistence to the Object store.

Using no_sync=true is best when prioritizing high-throughput writes over absolute durability.

• Default behavior (no_sync=false): Waits for data to be written to the Object store before acknowledging the write. Reduces the risk of data loss, but increases the latency of the response.
• With no_sync=true: Reduces write latency, but increases the risk of data loss in case of a crash before WAL persistence.

Immediate write using the HTTP API

The no_sync parameter controls when writes are acknowledged–for example:

curl "http://localhost:8181/api/v3/write_lp?db=sensors&precision=auto&no_sync=true" \
  --data-raw "home,room=Sunroom temp=96"

Immediate write using the influxdb3 CLI

The no_sync CLI option controls when writes are acknowledged–for example:

influxdb3 write --bucket=mydb --org=my_org --token=my-token --no-sync

Create a database or table

To create a database without writing data, use the create subcommand–for example:

influxdb3 create database mydb

To learn more about a subcommand, use the -h, --help flag:

influxdb3 create -h

Query data

InfluxDB 3 now supports native SQL for querying, in addition to InfluxQL, an SQL-like language customized for time series queries.

InfluxDB 3 Core limits query time ranges to 72 hours (both recent and historical) to ensure query performance. For more information about the 72-hour limitation, see the update on InfluxDB 3 Core’s 72-hour limitation.

The quickest way to get started querying is to use the influxdb3 CLI (which uses the Flight SQL API over HTTP2).

The query subcommand includes options to help ensure that the right database is queried with the correct permissions. Only the --database option is required, but depending on your specific setup, you may need to pass other options, such as host, port, and token.

Example: query “SHOW TABLES” on the servers database:

$ influxdb3 query --database=servers "SHOW TABLES"
+---------------+--------------------+--------------+------------+
| table_catalog | table_schema       | table_name   | table_type |
+---------------+--------------------+--------------+------------+
| public        | iox                | cpu          | BASE TABLE |
| public        | information_schema | tables       | VIEW       |
| public        | information_schema | views        | VIEW       |
| public        | information_schema | columns      | VIEW       |
| public        | information_schema | df_settings  | VIEW       |
| public        | information_schema | schemata     | VIEW       |
+---------------+--------------------+--------------+------------+

Example: query the cpu table, limiting to 10 rows:

$ influxdb3 query --database=servers "SELECT DISTINCT usage_percent, time FROM cpu LIMIT 10"
+---------------+---------------------+
| usage_percent | time                |
+---------------+---------------------+
| 63.4          | 2024-02-21T19:25:00 |
| 25.3          | 2024-02-21T19:06:40 |
| 26.5          | 2024-02-21T19:31:40 |
| 70.1          | 2024-02-21T19:03:20 |
| 83.7          | 2024-02-21T19:30:00 |
| 55.2          | 2024-02-21T19:00:00 |
| 80.5          | 2024-02-21T19:05:00 |
| 60.2          | 2024-02-21T19:33:20 |
| 20.5          | 2024-02-21T18:58:20 |
| 85.2          | 2024-02-21T19:28:20 |
+---------------+---------------------+

Querying using the CLI for InfluxQL

InfluxQL is an SQL-like language developed by InfluxData with specific features tailored for leveraging and working with InfluxDB. It’s compatible with all versions of InfluxDB, making it a good choice for interoperability across different InfluxDB installations.

To query using InfluxQL, enter the influxdb3 query subcommand and specify influxql in the language option–for example:

influxdb3 query --database=servers --language=influxql "SELECT DISTINCT usage_percent FROM cpu WHERE time >= now() - 1d"

Query using the API

InfluxDB 3 supports Flight (gRPC) APIs and an HTTP API. To query your database using the HTTP API, send a request to the /api/v3/query_sql or /api/v3/query_influxql endpoints. In the request, specify the database name in the db parameter and a query in the q parameter. You can pass parameters in the query string or inside a JSON object.

Use the format parameter to specify the response format: pretty, jsonl, parquet, csv, and json. Default is json.

Example: Query passing URL-encoded parameters

The following example sends an HTTP GET request with a URL-encoded SQL query:

curl -v "http://localhost:8181/api/v3/query_sql?db=servers&q=select+*+from+cpu+limit+5"

Example: Query passing JSON parameters

The following example sends an HTTP POST request with parameters in a JSON payload:

curl http://localhost:8181/api/v3/query_sql \
  --data '{"db": "server", "q": "select * from cpu limit 5"}'

Query using the Python client

Use the InfluxDB 3 Python library to interact with the database and integrate with your application. We recommend installing the required packages in a Python virtual environment for your specific project.

To get started, install the influxdb3-python package.

pip install influxdb3-python

From here, you can connect to your database with the client library using just the host and **database name:

from influxdb_client_3 import InfluxDBClient3

client = InfluxDBClient3(
    host='http://localhost:8181',
    database='servers'
)

The following example shows how to query using SQL, and then use PyArrow to explore the schema and process results:

from influxdb_client_3 import InfluxDBClient3

client = InfluxDBClient3(
    host='http://localhost:8181',

    database='servers'
)

# Execute the query and return an Arrow table
table = client.query(
    query="SELECT * FROM cpu LIMIT 10",
    language="sql"
)

print("\n#### View Schema information\n")
print(table.schema)

print("\n#### Use PyArrow to read the specified columns\n")
print(table.column('usage_active'))
print(table.select(['host', 'usage_active']))
print(table.select(['time', 'host', 'usage_active']))

print("\n#### Use PyArrow compute functions to aggregate data\n")
print(table.group_by('host').aggregate([]))
print(table.group_by('cpu').aggregate([('time_system', 'mean')]))

For more information about the Python client library, see the influxdb3-python repository in GitHub.

Last values cache

InfluxDB 3 Core supports a last-n values cache which stores the last N values in a series or column hierarchy in memory. This gives the database the ability to answer these kinds of queries in under 10 milliseconds. You can use the influxdb3 CLI to create a last value cache.

Usage: $ influxdb3 create last_cache [OPTIONS] -d <DATABASE_NAME> -t <TABLE> [CACHE_NAME]

Options:
  -h, --host <HOST_URL>                URL of the running InfluxDB 3 Core server [env: INFLUXDB3_HOST_URL=] 
  -d, --database <DATABASE_NAME>       The database to run the query against [env: INFLUXDB3_DATABASE_NAME=]
      --token <AUTH_TOKEN>             The token for authentication [env: INFLUXDB3_AUTH_TOKEN=]
  -t, --table <TABLE>                  The table for which the cache is created
      --key-columns <KEY_COLUMNS>      Columns used as keys in the cache
      --value-columns <VALUE_COLUMNS>  Columns to store as values in the cache
      --count <COUNT>                  Number of entries per unique key:column 
      --ttl <TTL>                      The time-to-live for entries (seconds)
      --help                           Print help information

You can create a last values cache per time series, but be mindful of high cardinality tables that could take excessive memory.

An example of creating this cache in use:

influxdb3 create last_cache --database=servers --table=cpu --key-columns=host,application --value-columns=usage_percent,status --count=5 cpuCache

Query a Last values cache

To leverage the LVC, call it using the last_cache() function in your query–for example:

influxdb3 query --database=servers "SELECT * FROM last_cache('cpu', 'cpuCache') WHERE host = 'Bravo;"

Deleting a Last values cache

To remove a Last values cache, use the following command:

influxdb3 delete last_cache [OPTIONS] -d <DATABASE_NAME> -t <TABLE> --cache-name <CACHE_NAME>

Options:
  -h, --host <HOST_URL>          Host URL of the running InfluxDB 3 server
  -d, --database <DATABASE_NAME> The database to run the query against          
      --token <AUTH_TOKEN>       The token for authentication   
  -t, --table <TABLE>            The table for which the cache is being deleted
  -n, --cache-name <CACHE_NAME>  The name of the cache being deleted
      --help                     Print help information

Distinct values cache

Similar to the Last values cache, the database can cache in RAM the distinct values for a single column in a table or a heirarchy of columns. This is useful for fast metadata lookups, which can return in under 30 milliseconds. Many of the options are similar to the last value cache. See the CLI output for more information:

influxdb3 create distinct_cache -h

Python plugins and the Processing engine

The InfluxDB 3 Processing engine is an embedded Python VM for running code inside the database to process and transform data.

To activate the Processing engine, pass the --plugin-dir <PLUGIN_DIR> option when starting the InfluxDB 3 Core server. PLUGIN_DIR is your filesystem location for storing plugin files for the Processing engine to run.

Plugin

A plugin is a Python function that has a signature compatible with a Processing engine trigger.

Trigger

When you create a trigger, you specify a plugin, a database, optional arguments, and a trigger-spec, which defines when the plugin is executed and what data it receives.

Trigger types

InfluxDB 3 provides the following types of triggers, each with specific trigger-specs:

• On WAL flush: Sends a batch of written data (for a specific table or all tables) to a plugin (by default, every second).
• On Schedule: Executes a plugin on a user-configured schedule (using a crontab or a duration); useful for data collection and deadman monitoring.
• On Request: Binds a plugin to a custom HTTP API endpoint at /api/v3/engine/<ENDPOINT>. The plugin receives the HTTP request headers and content, and can then parse, process, and send the data into the database or to third-party services.

Test, create, and trigger plugin code

Example: Python plugin for WAL flush

# This is the basic structure for Python plugin code that runs in the
# InfluxDB 3 Processing engine.

# When creating a trigger, you can provide runtime arguments to your plugin,
# allowing you to write generic code that uses variables such as monitoring
thresholds, environment variables, and host names.
#
# Use the following exact signature to define a function for the WAL flush
# trigger.
# When you create a trigger for a WAL flush plugin, you specify the database
# and tables that the plugin receives written data from on every WAL flush
# (default is once per second).
def process_writes(influxdb3_local, table_batches, args=None):
    # here you can see logging. for now this won't do anything, but soon 
    # we'll capture this so you can query it from system tables
    if args and "arg1" in args:
        influxdb3_local.info("arg1: " + args["arg1"])

    # here we're using arguments provided at the time the trigger was set up 
    # to feed into paramters that we'll put into a query
    query_params = {"host": "foo"}
    # here's an example of executing a parameterized query. Only SQL is supported. 
    # It will query the database that the trigger is attached to by default. We'll 
    # soon have support for querying other DBs.
    query_result = influxdb3_local.query("SELECT * FROM cpu where host = '$host'", query_params)
    # the result is a list of Dict that have the column name as key and value as 
    # value. If you run the WAL test plugin with your plugin against a DB that 
    # you've written data into, you'll be able to see some results
    influxdb3_local.info("query result: " + str(query_result))

    # this is the data that is sent when the WAL is flushed of writes the server 
    # received for the DB or table of interest. One batch for each table (will 
    # only be one if triggered on a single table)
    for table_batch in table_batches:
        # here you can see that the table_name is available.
        influxdb3_local.info("table: " + table_batch["table_name"])

        # example to skip the table we're later writing data into
        if table_batch["table_name"] == "some_table":
            continue

        # and then the individual rows, which are Dict with keys of the column names and values
        for row in table_batch["rows"]:
            influxdb3_local.info("row: " + str(row))

    # this shows building a line of LP to write back to the database. tags must go first and 
    # their order is important and must always be the same for each individual table. Then 
    # fields and lastly an optional time, which you can see in the next example below
    line = LineBuilder("some_table")\
        .tag("tag1", "tag1_value")\
        .tag("tag2", "tag2_value")\
        .int64_field("field1", 1)\
        .float64_field("field2", 2.0)\
        .string_field("field3", "number three")
    
    # this writes it back (it actually just buffers it until the completion of this function
    # at which point it will write everything back that you put in)
    influxdb3_local.write(line)

    # here's another example, but with us setting a nanosecond timestamp at the end
    other_line = LineBuilder("other_table")
    other_line.int64_field("other_field", 1)
    other_line.float64_field("other_field2", 3.14)
    other_line.time_ns(1302)

    # and you can see that we can write to any DB in the server
    influxdb3_local.write_to_db("mytestdb", other_line)

    # just some log output as an example
    influxdb3_local.info("done")

Test a plugin on the server

Test your InfluxDB 3 plugin safely without affecting written data. During a plugin test:

• A query executed by the plugin queries against the server you send the request to.
• Writes aren’t sent to the server but are returned to you.

To test a plugin, do the following:

1. Create a plugin directory–for example, /path/to/.influxdb/plugins

2. Start the InfluxDB server and include the --plugin-dir <PATH> option.

3. Save the preceding example code to a plugin file inside of the plugin directory. If you haven’t yet written data to the table in the example, comment out the lines where it queries.

4. To run the test, enter the following command with the following options:

   • --lp or --file: The line protocol to test
   • Optional: --input-arguments: A comma-delimited list of <KEY>=<VALUE> arguments for your plugin code

   influxdb3 test wal_plugin \
     --lp <INPUT_LINE_PROTOCOL> \
     --input-arguments "arg1=foo,arg2=bar"
     --database <DATABASE_NAME> \
     <PLUGIN_FILENAME> 
   

   • Copy
   • Fill window

The command runs the plugin code with the test data, yields the data to the plugin code, and then responds with the plugin result. You can quickly see how the plugin behaves, what data it would have written to the database, and any errors. You can then edit your Python code in the plugins directory, and rerun the test. The server reloads the file for every request to the test API.

For more information, see influxdb3 test wal_plugin or run influxdb3 test wal_plugin -h.

With the plugin code inside the server plugin directory, and a successful test, you’re ready to create a plugin and a trigger to run on the server.

Example: Test, create, and run a plugin

The following example shows how to test a plugin, and then create the plugin and trigger:

# Test and create a plugin
# Requires:
#   - A database named `mydb` with a table named `foo`
#   - A Python plugin file named `test.py`
# Test a plugin
influxdb3 test wal_plugin \
  --lp="my_measure,tag1=asdf f1=1.0 123" \
  -d mydb \
  --input-arguments="arg1=hello,arg2=world" \
  test.py

# Create a trigger that runs the plugin
influxdb3 create trigger \
  -d mydb \
  --plugin=test_plugin \
  --trigger-spec="table:foo" \
  --trigger-arguments="arg1=hello,arg2=world" \
  trigger1

After you have created a plugin and trigger, enter the following command to enable the trigger and have it run the plugin as you write data:

influxdb3 enable trigger --database mydb trigger1 

For more information, see Python plugins and the Processing engine.