Skip to content

Latest commit

 

History

History
119 lines (65 loc) · 5.01 KB

README.md

File metadata and controls

119 lines (65 loc) · 5.01 KB

Generating Perfrunner Docker Images

Deploying with docker-compose

Note: this isn't recommended for production deployments. However, this is useful for development where you don't wish to install the client dependencies on your machine, or have an incompatible OS.

Running perfrunner with docker-compose allows the perfrunner repo to be mounted as a volume, so live changes you make inside the repo, on your host machine, will be reflected in the container. Anything outside the repo (for example, system packages), will not be mirrored between container and host. This gives you a self-contained development environment.

Requirements

Setup

First create a local testing network. This provides isolation and you only have to do it once:

docker network create local-perf-testing

Keep the name the same, perfrunner docker compose is setup to find a network with this name. Then, to build, install perfrunner requirements, and gain a shell in the container:

make docker-compose CONTAINER_PASSWORD=${password}

Where ${password} is the same as that supplied in the credentials section of your spec file.

Simply disconnect with exit when done.

Other useful commands include:

  • Exec back into the container:

      docker exec -it perfrunner /bin/bash

  • Kill the perfrunner container:

      docker kill perfrunner

  • Restart the perfrunner container:

      docker start perfrunner

Installing and Connecting to Couchbase Server

Installing Couchbase server using Docker

Follow the instructions from the documentations on how to install a couchbase server locally using docker. Remember to add --network local-perf-testing in your docker run command when starting the couchbase server.

Configuring Server Containers

By installing and enabling the SSH server on the server nodes, you ensure that Perfrunner can securely interact with the Couchbase cluster, execute commands, and perform data operations seamlessly throughout the performance testing phases.

Install SSH Server:

apt-get update
apt-get install -y openssh-server

Once installed, start the SSH service using service ssh start. These commands ensure that the SSH server is installed and running within the container, allowing secure remote connections.

Permitting Root Login

Set Password:

Use the passwd command to set a password for the SSH server. This command prompts you to enter and confirm the new password.

Configure SSH:

Use your preferred editor to open the SSH configuration file located at /etc/ssh/sshd_config. Within the SSH configuration file, locate the PermitRootLogin line and uncomment it if it exists or add it if it's not present. Set the value to yes to allow root login via SSH. Save and exit the editor.

Restart SSH Service:

Finally, restart the SSH service using service ssh restart to apply the changes made to the SSH server configuration.

Preparing a local docker cluster spec file

To run a perfrunner test in a container, you can run commands as normal, eg:

env/bin/perfrunner -c clusters/cluster_spec.spec -t tests/comp_bucket_20M.test

Note: Make sure the cluster specified by -c is free and there are no any running jobs at the moment using the cluster.

To prepare a local cluster spec for testing, copy any example cluster available in clusters/ directory and name it say clusters/local.spec. Modify the following two sections, clusters and clients as in the example below:

[clusters]
local =
    <server-ip>:kv

[clients]
hosts =
    localhost

The client hosts should be localhost or 127.0.0.1. The server IP can be obtained by inspecting your Couchbase server docker container and look for IPAddress. For instance, if my server container name is couchbase:

    docker inspect couchbase | grep IPAddress -

Generating docker images for Kubernetes workers

To generate docker images available at docker hub, run the following commands:

  • Make sure the image is built for linux/amd64. This is what we use for our client machines. You can skip this step if running on linux/amd64 platform or building for local testing.

      export DOCKER_DEFAULT_PLATFORM=linux/amd64

  • Optionaly also specify PYTHON_VERSIONS and PYENV_VERSION build arguments (with --build-arg) to specify which Python versions and Pyenv version to install in the image.

  • Build perfrunner image:

      docker build --target perfrunner docker/

  • Tag the image appropriately

      docker tag <generated-image-tag> perflab/perfrunner

    The generated image tag will be displaed at the end of the previous build step.

This will tag the new image as @latest. You can use your personal hub to build experimental builds. Or simply:

    make docker