influxdb-cxx

InfluxDB C++ client library

• Batch write
• Data exploration
• Supported transports
  • HTTP/HTTPS
  • UDP
  • Unix datagram socket
  • TCP

Fork of the unmaintained awegrzyn/influxdb-cxx project.

Installation

Build requirements

• CMake 3.12+
• C++20 compiler

Dependencies

• cpr (required)ⁱ⁾
• boost 1.78+ (optional – see Transports)

ⁱ⁾ cpr needs to provide CMake support; some systems need to call ldconfig after .so installation.

Generic

mkdir build && cd build
cmake ..
sudo make install

Quick start

Include in CMake project

The InfluxDB library is exported as target InfluxData::InfluxDB.

project(example)

find_package(InfluxDB)

add_executable(example-influx main.cpp)
target_link_libraries(example-influx PRIVATE InfluxData::InfluxDB)

This target is also provided when the project is included as a subdirectory.

project(example)
add_subdirectory(influxdb-cxx)
add_executable(example-influx main.cpp)
target_link_libraries(example-influx PRIVATE InfluxData::InfluxDB)

Basic write

// Provide complete URI
auto influxdb = influxdb::InfluxDBFactory::Get("http://localhost:8086?db=test");
influxdb->write(influxdb::Point{"test"}
  .addField("value", 10)
  .addTag("host", "localhost")
);

Batch write

auto influxdb = influxdb::InfluxDBFactory::Get("http://localhost:8086?db=test");
// Write batches of 100 points
influxdb->batchOf(100);

for (;;) {
  influxdb->write(influxdb::Point{"test"}.addField("value", 10));
}

Note:

When batch write is enabled, call flushBatch() to flush pending batches. This is of particular importance to ensure all points are written prior to destruction.

auto influxdb = influxdb::InfluxDBFactory::Get("http://localhost:8086?db=test");
influxdb->batchOf(3);
influxdb->write(influxdb::Point{"test"}.addField("value", 1));
influxdb->write(influxdb::Point{"test"}.addField("value", 2));

// Flush batches, both points are written
influxdb->flushBatch();

Query

// Available over HTTP only
auto influxdb = influxdb::InfluxDBFactory::Get("http://localhost:8086?db=test");
/// Pass an IFQL to get list of points
std::vector<influxdb::Point> points = influxdb->query("SELECT * FROM test");

Execute cmd

auto influxdb = influxdb::InfluxDBFactory::Get("http://localhost:8086?db=test");

// Execute a command and receive it's response
const auto response = influxdb->execute("SHOW DATABASES");

Transports

Supported transports:

Name	Dependency	URI protocol	Sample URI
HTTP	cpri)	http/https	http://localhost:8086?db=<db>
TCP	boost	tcp	tcp://localhost:8094
UDP	boost	udp	udp://localhost:8094
Unix socket	boost	unix	unix:///tmp/telegraf.sock

ⁱ⁾ boost is needed to support queries.

Configuration by URI

An underlying transport is configurable by passing an URI. [protocol] determines the actual transport type:

auto influxdb = influxdb::InfluxDBFactory::Get("http://localhost:8086?db=test");

URI Format:

[protocol]://[username:password@]host:port[?db=database]

# Auth token:
[protocol]://[token@]host:port[?db=database]

Additional transport configuration (HTTP only)

The HTTP transport supports additional configurations beyond the limited URI parameters:

auto influxdb = InfluxDBBuilder::http("http://localhost:8086?db=test")
                    .setTimeout(std::chrono::seconds{20})
                    .setAuthToken("<token>")
                    .connect();

InfluxDB v2.x compatibility

The support for InfluxDB v2.x is limited at the moment. It's possible to use the v1.x compatibility backend though.

Please visit InfluxDB 1.x compatibility API docs for more information.

To create a v1.x compatible user (as described here):

influx v1 auth create --read-bucket ${BUCKET_ID} --write-bucket ${BUCKET_ID} --username ${USERNAME} --password ${PASSWORD}

Thread safety

This library is not thread-safe. Using it in a multi-threaded environment without proper synchronization mechanisms may lead to unexpected behavior and data corruption.