Install Vector From Source

This page covers installing Vector from source using the native toolchain for the host.

Vector can also be compiled to a static binary for Linux for x86_64, ARM64, and ARMv7 architectures. See compiling using Docker for details.

We recommend installing Vector through a supported container platform, package manager, or pre-built archive, if possible. These handle permissions, directory creation, and other intricacies covered in the Next Steps section.

Installation

1. Install Rust

   curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh -s -- -y --default-toolchain stable

2. Install C++ toolchain

   Install C and C++ compilers (GCC or Clang) and GNU make if they are not pre-installed on your system.

3. Create the vector directory

   mkdir vector

4. Download Vector's Source

   • Latest (0.6.0)
   • Master

   mkdir -p vector && \
     curl -sSfL --proto '=https' --tlsv1.2 https://api.github.com/repos/timberio/vector/tarball/v0.6.X | \
     tar xzf - -C vector --strip-components=1

5. Change into the vector directory

   cd vector

6. Compile Vector

   [FEATURES="<flag1>,<flag2>,..."] make build

   The FEATURES environment variable is optional. You can override the default features with this variable. See feature flags for more info.

   When finished, the vector binary will be placed in target/<target>/release/vector. For example, if you are building Vector on your Mac, your target triple is x86_64-apple-darwin, and the Vector binary will be located at target/x86_64-apple-darwin/release/vector.

7. Start Vector

   Finally, start vector:

   target/<target>/release/vector --config config/vector.toml

Next Steps

Configuring

The Vector configuration file is located at:

config/vector.toml

A full spec is located at config/vector.spec.toml and examples are located in config/vector/examples/*. You can learn more about configuring Vector in the Configuration section.

Data Directory

We highly recommend creating a data directory that Vector can use:

mkdir /var/lib/vector

Make sure that this directory is writable by the vector process.

Vector offers a global data_dir option that you can use to specify the path of your directory.

vector.toml

data_dir = "/var/lib/vector" # default

Service Managers

Vector archives ship with service files in case you need them:

Init.d

To install Vector into Init.d run:

cp -av distribution/init.d/vector /etc/init.d

Systemd

To install Vector into Systemd run:

cp -av distribution/systemd/vector.service /etc/systemd/system

Updating

Simply follow the same installation instructions above.

How It Works

Feature Flags

The following feature flags are supported via the FEATURES env var when executing make build:

[FEATURES="<flag1>,<flag2>,..."] make build

There are three meta-features which can be used when compiling for the corresponding targets. If no features are specified, then the default one is used.

Feature	Description	Enabled by default
default	Default set of features for *-unknown-linux-gnu and *-apple-darwin targets.
default-musl	Default set of features for *-unknown-linux-musl targets. Requires cmake and perl as build dependencies.
default-msvc	Default set of features for *-pc-windows-msvc targets. Requires cmake and perl as build dependencies.

Alternatively, for finer control, it is possible to use specific features from the list below:

Feature	Description	Included in default feature
unix	Enables features that require cfg(unix) to be present on the platform, namely support for Unix domain sockets in docker source and jemalloc instead of the default memory allocator.
vendored	Forces vendoring of OpenSSL and ZLib dependencies instead of using their versions installed in the system. Requires perl as a build dependency.
leveldb-plain	Enables support for disk buffers using vendored LevelDB.
leveldb-cmake	The same as leveldb-plain, but is more portable. Requires cmake as a build dependency. Use it in case of compilation issues with leveldb-plain.
rdkafka-plain	Enables vendored librdkafka dependency, which is required for kafka source and kafka sink.
rdkafka-cmake	The same as rdkafka-plain, but is more portable. Requires cmake as a build dependency. Use it in case of compilation issues with rdkafka-plain.

Compiling using Docker

It is possible to build statically linked binaries of Vector for Linux using Docker.

In this case the dependencies listed in the previous section are not needed, as all of them would be automatically pulled by Docker.

Building steps:

1. Download Vector's Source

   • Latest (0.6.0)
   • Master

   mkdir -p vector && \
     curl -sSfL --proto '=https' --tlsv1.2 https://api.github.com/repos/timberio/vector/tarball/v0.6.X | \
     tar xzf - -C vector --strip-components=1

2. Build Vector using Docker

   • Linux (x86_64)
   • Linux (ARM64)
   • Linux (ARMv7)

   PASS_FEATURES=default-musl ./scripts/docker-run.sh builder-x86_64-unknown-linux-musl make build

   The command above builds a Docker image with Rust toolchain for a Linux target for the corresponding architecture using musl as the C library, then starts a container from this image, and then builds inside the Container. The target binary is located in target/<target triple>/release/vector like in the previous case.