InfluxDB Line Protocol Overview

QuestDB implements the InfluxDB Line Protocol to ingest data.

The InfluxDB Line Protocol is for data ingestion only.

For building queries, see the Query & SQL Overview.

Each ILP client library also has its own language-specific documentation set.

This supporting document thus provides an overview to aid in client selection and initial configuration:

  1. Client libraries
  2. Server-Side Configuration
  3. Transport Selection
  4. Client-Side Configuration
  5. Error handling
  6. Authentication
  7. Table and Column Auto-creation
  8. Timestamp Column Name
  9. HTTP Transaction semantics
  10. Exactly-once delivery
  11. Health Check

Client libraries

The quickest way to get started is to select your library of choice.

From there, its documentation will carry you through to implementation.

Client libraries are available for several languages:

If you'd like more context on ILP overall, please continue reading.

Enable or disable ILP

If going over HTTP, ILP will use shared HTTP port 9000 (default) if the following is set in server.conf:

line.http.enabled=true

Server-Side Configuration

The HTTP receiver configuration can be completely customized using QuestDB configuration keys for ILP.

Configure the thread pools, buffer and queue sizes, receiver IP address and port, load balancing, and more.

For more guidance in how to tune QuestDB, see capacity planning.

Transport selection

The ILP protocol in QuestDB supports the following transport options:

  • HTTP (default port 9000)
  • TCP (default port 9009)

On QuestDB Enterprise HTTPS and TCPS are also available.

The HTTP(s) transport is recommended for most use cases. It provides feedback on errors, automatically retries failed requests, and is easier to configure. The TCP(s) transport is kept for compatibility with older QuestDB versions. It has limited error feedback, no automatic retries, and requires manual handling of connection failures. However, while HTTP is recommended, TCP has slightly lower overhead than HTTP and may be useful in high-throughput scenarios in high-latency networks.

Client-Side Configuration

Clients connect to a QuestDB using ILP via a configuration string. Configuration strings combine a set of key/value pairs.

The standard configuration string pattern is:


schema::key1=value1;key2=value2;key3=value3;

It is made up of the following parts:

  • Schema: One of the specified schemas in the core parameters section below
  • Key=Value: Each key-value pair sets a specific parameter for the client
  • Terminating semicolon: A semicolon must follow the last key-value pair

Client parameters

Below is a list of common parameters that ILP clients will accept.

These params facilitate connection to QuestDB's ILP server and define client-specific behaviors.

Some are shared across all clients, while some are client specific. Refer to the clients documentation for details.

warning

Any parameters tagged as SENSITIVE must be handled with care.

Exposing these values may expose your database to bad actors.

Core parameters

  • schema: Specifies the transport method, with support for: http, https, tcp & tcps
  • addr: The address and port of the QuestDB server, as in localhost:9000.

HTTP Parameters

  • password (SENSITIVE): Password for HTTP Basic Authentication.
  • request_min_throughput: Expected throughput for network send to the database server, in bytes.
    • Defaults to 100 KiB/s
    • Used to calculate a dynamic timeout for the request, so that larger requests do not prematurely timeout.
  • request_timeout: Base timeout for HTTP requests to the database, in milliseconds.
    • Defaults to 10 seconds.
  • retry_timeout: Maximum allowed time for client to attempt retries, in milliseconds.
    • Defaults to 10 seconds.
    • Not all errors are retriable.
  • token (SENSITIVE): Bearer token for HTTP Token authentication.
    • Open source HTTP users are unable to generate tokens. For TCP token auth, see the below section.
  • username: Username for HTTP Basic Authentication.

TCP Parameters

note

These parameters are only useful when using ILP over TCP with authentication enabled. Most users should use ILP over HTTP. These parameters are listed for completeness and for users who have specific requirements.

See the Authentication section below for configuration.

  • auth_timeout: Timeout for TCP authentication with QuestDB server, in milliseconds.
    • Default 15 seconds.
  • token (SENSITIVE): TCP Authentication d parameter.
    • token_x (SENSITIVE): TCP Authentication x parameter.
      • Used in C/C++/Rust/Python clients.
    • token_y (SENSITIVE): TCP Authentication y parameter.
      • Used in C/C++/Rust/Python clients.
  • username: Username for TCP authentication.

Auto-flushing behavior

  • auto_flush: Enable or disable automatic flushing (on/off).

    • Default is “on” for clients that support auto-flushing (all except C, C++ & Rust).
  • auto_flush_bytes Auto-flushing is triggered above this buffer size.

    • Disabled by default.
  • auto_flush_interval: Auto-flushing is triggered after this time period has elapsed since the last flush, in milliseconds.

    • Defaults to 1 second
    • This is not a periodic timer - it will only be checked on the next row creation.
  • auto_flush_rows: Auto-flushing is triggered above this row count.

    • Defaults to 75,000 for HTTP, and 600 for TCP.
    • If set, this implies “auto_flush=on”.

Buffer configuration

  • init_buf_size: Set the initial (but growable) size of the buffer in bytes.
    • Defaults to 64 KiB.
  • max_buf_size: Sets the growth limit of the buffer in bytes.
    • Defaults to 100 MiB.
    • Clients will error if this is exceeded.
  • max_name_len: The maximum alloable number of UTF-8 bytes in the table or column names.
    • Defaults to 127.
    • Related to length limits for filenames on the user's host OS.

TLS configuration

QuestDB Enterprise only.

  • tls_verify: Toggle verification of TLS certificates. Default is on.
  • tls_roots: Specify the source of bundled TLS certificates.
    • The defaults and possible param values are client-specific.
      • In Rust and Python this might be “webpki”, “os-certs” or a path to a “pem” file.
      • In Java this might be a path to a “jks” trust store.
      • tls_roots_password Password to a configured tls_roots if any.
        • Passwords are sensitive! Manage appropriately.
  • tls_ca: Path to single certificate authourity, not supported on all clients.
    • Java for instance would apply tls_roots=/path/to/Java/key/store

Network configuration

  • bind_interface: Optionally, specify the local network interface for outbound connections. Useful if you have multiple interfaces or an accelerated network interface (e.g. Solarflare)
    • Not to be confused with the QuestDB port in the addr param.

Error handling

The HTTP transport supports automatic retries for failed requests deemed recoverable. Recoverable errors include network errors, some server errors, and timeouts, while non-recoverable errors encompass invalid data, authentication errors, and other client-side errors.

Retrying is particularly beneficial during network issues or when the server is temporarily unavailable. The retrying behavior can be configured through the retry_timeout configuration option or, in some clients, via their API. The client continues to retry recoverable errors until they either succeed or the specified timeout is reached.

The TCP transport lacks support for error propagation from the server. In such cases, the server merely closes the connection upon encountering an error. Consequently, the client receives no additional error information from the server. This limitation significantly contributes to the preference for HTTP transport over TCP transport.

Authentication

note

Using QuestDB Enterprise?

Skip to advanced security features instead, which provides holistic security out-of-the-box.

InfluxDB Line Protocol supports authentication via HTTP Basic Authentication, using the HTTP Parameters, or via token when using the TCP transport, using the TCP Parameters.

A similar pattern is used across all client libraries. If you want to use a TCP token, you need to configure your QuestDB server. This document will break down and demonstrate the configuration keys and core configuration options.

Once a client has been selected and configured, resume from your language client documentation.

TCP token authentication setup

Create d, x & y tokens for client usage.

Prerequisites
  • jose: C-language implementation of Javascript Object Signing and Encryption. Generates tokens.
  • jq: For pretty JSON output.
brew install jose
brew install jq
Server configuration

Next, create an authentication file.

Only elliptic curve (P-256) are supported (key type ec-p-256-sha256):

testUser1 ec-p-256-sha256 fLKYEaoEb9lrn3nkwLDA-M_xnuFOdSt9y0Z7_vWSHLU Dt5tbS1dEDMSYfym3fgMv0B99szno-dFc1rYF9t0aac
# [key/user id] [key type] {keyX keyY}

Generate an authentication file using the jose utility:

jose jwk gen -i '{"alg":"ES256", "kid": "testUser1"}' -o /var/lib/questdb/conf/full_auth.json

KID=$(cat /var/lib/questdb/conf/full_auth.json | jq -r '.kid')
X=$(cat /var/lib/questdb/conf/full_auth.json | jq -r '.x')
Y=$(cat /var/lib/questdb/conf/full_auth.json | jq -r '.y')

echo "$KID ec-p-256-sha256 $X $Y" | tee /var/lib/questdb/conf/auth.txt

Once created, reference it in the server configuration:

/path/to/server.conf
line.tcp.auth.db.path=conf/auth.txt
Client keys

For the server configuration above, the corresponding JSON Web Key must be stored on the clients' side.

When sending a fully-composed JWK, it will have the following keys:

{
"kty": "EC",
"d": "5UjEMuA0Pj5pjK8a-fa24dyIf-Es5mYny3oE_Wmus48",
"crv": "P-256",
"kid": "testUser1",
"x": "fLKYEaoEb9lrn3nkwLDA-M_xnuFOdSt9y0Z7_vWSHLU",
"y": "Dt5tbS1dEDMSYfym3fgMv0B99szno-dFc1rYF9t0aac"
}

The d, x and y parameters generate the public key.

For example, the Python client would be configured as outlined in the Python docs.

Table and column auto-creation

When sending data to a table that does not exist, the server will create the table automatically. This also applies to columns that do not exist. The server will use the first row of data to determine the column types.

If the table already exists, the server will validate that the columns match the existing table. If the columns do not match, the server will return a non-recoverable error which, when using the HTTP/HTTPS transport, is propagated to the client.

You can avoid table and/or column auto-creation by setting the line.auto.create.new.columns and line.auto.create.new.tablesconfiguration parameters to false.

If you're using QuestDB Enterprise, you must grant further permissions to the authenticated user:

CREATE SERVICE ACCOUNT ingest_user; -- creates a service account to be used by a client
GRANT ilp, create table TO ingest_user; -- grants permissions to ingest data and create tables
GRANT add column, insert ON all tables TO ingest_user; -- grants permissions to add columns and insert data to all tables
-- OR
GRANT add column, insert ON table1, table2 TO ingest_user; -- grants permissions to add columns and insert data to specific tables

Read more setup details in the Enterprise quickstart and the role-based access control guides.

Timestamp Column Name

QuestDB's underlying ILP protocol sends timestamps to QuestDB without a name.

If your table has been created beforehand, the designated timestamp will be correctly assigned based on the payload sent bt the client. But if your table does not exist, it will be automatically created and the timestamp column will be named timestamp. To use a custom name, say my_ts, pre-create the table with the desired timestamp column name.

To do so, issue a CREATE TABLE statement to create the table in advance:

Creating a timestamp named my_ts
CREATE TABLE IF NOT EXISTS 'trades' (
symbol SYMBOL capacity 256 CACHE,
side SYMBOL capacity 256 CACHE,
price DOUBLE,
amount DOUBLE,
my_ts TIMESTAMP
) timestamp (my_ts) PARTITION BY DAY WAL;

You can use the CREATE TABLE IF NOT EXISTS construct to make sure the table is created, but without raising an error if the table already exists.

HTTP transaction semantics

The TCP endpoint does not support transactions. The HTTP ILP endpoint treats every requests as an individual transaction, so long as it contains rows for a single table.

As of writing, the HTTP endpoint does not provide full transactionality in all cases.

Specifically:

  • If an HTTP request contains data for two tables and the final commit fails for the second table, the data for the first table will still be committed. This is a deviation from full transactionality, where a failure in any part of the transaction would result in the entire transaction being rolled back. If data transactionality is important for you, the best practice is to make sure you flush data to the server in batches that contain rows for a single table.

  • Even when you are sending data to a single table, when dynamically adding new columns to a table, an implicit commit occurs each time a new column is added. If the request is aborted or has parse errors, no data will be inserted into the corresponding table, but the new column will be added and will not be rolled back.

  • Some clients have built-in support for controlling transactions. These APIs help to comply with the single-table-per-request pre-requisite for HTTP transactions, but they don't control if new columns are being added.

  • As of writing, if you want to make sure you have data transactionality and schema/metadata transactionality, you should disable line.auto.create.new.columns and line.auto.create.new.tables on your configuration. Be aware that if you do this, you will not have dynamic schema capabilities and you will need to create each table and column before you try to ingest data, via CREATE TABLE and/or ALTER TABLE ADD COLUMN SQL statements.

Exactly-once delivery vs at-least-once delivery

The retrying behavior of the HTTP transport can lead to some data being sent to the server more than once.

Example: Client sends a batch to the server, the server receives the batch, processes it, but fails to send a response back to the client due to a network error. The client will retry sending the batch to the server. This means the server will receive the batch again and process it again. This can lead to duplicated rows in the server.

The are two ways to mitigate this issue:

  • Use QuestDB deduplication feature to remove duplicated rows. QuestDB server can detect and remove duplicated rows automatically, resulting in exactly-once processing. This is recommended when using the HTTP transport with retrying enabled.
  • Disable retrying by setting retry_timeout to 0. This will make the client send the batch only once, failed requests will not be retried and the client will receive an error. This effectively turns the client into an at-most-once delivery.

Health Check

To monitor your active connection, there is a ping endpoint:

curl -I http://localhost:9000/ping

Returns (pong!):

HTTP/1.1 204 OK
Server: questDB/1.0
Date: Fri, 2 Feb 2024 17:09:38 GMT
Transfer-Encoding: chunked
Content-Type: text/plain; charset=utf-8
X-Influxdb-Version: v2.7.4

Determine whether an instance is active and confirm the version of InfluxDB Line Protocol with which you are interacting.