Deployments on HUATUO Documentation

Docker

Sun, 11 Jan 2026 00:00:00 +0000

Image Download

Image repository: https://hub.docker.com/r/huatuo/huatuo-bamai/tags

Start a container with Docker

docker run --detach \
  --name huatuo-bamai \
  --restart unless-stopped \
  --privileged \
  --pid=host \
  --cgroupns=host \
  --network=host \
  --cpus=2 \
  --memory=2g \
  --volume /sys:/sys \
  --volume /proc:/proc \
  --volume /run:/run \
  huatuo/huatuo-bamai:latest

Note: The built-in default configuration does not connect to kubelet or Elasticsearch.

Limit CPU and memory in production to isolate abnormal collection workloads. Docker manages the container cgroup; Huatuo does not create its own cgroup by default, so do not pass --enable-cgroup in a Docker deployment.

Kubernetes

Sun, 11 Jan 2026 00:00:00 +0000

This document describes how to deploy the Huatuo collector to a Kubernetes cluster using a DaemonSet.

1. Kubernetes Manifest Deployment

1.1 Download the configuration file

curl -L -o huatuo-bamai.conf https://github.com/ccfos/huatuo/raw/main/huatuo-bamai.conf

1.2 Modify the configuration file

Modify the configuration file for the deployment environment. For example, configure the storage backend and the method used to obtain Pod information. See the Configuration Guide for details.

1.3 Create the ConfigMap

kubectl create configmap huatuo-bamai-config \
  --namespace default \
  --from-file=./huatuo-bamai.conf \
  --dry-run=client -o yaml |
kubectl apply -f -

1.4 Deploy the collector

Download the DaemonSet manifest:

Bare-Metal

Sun, 11 Jan 2026 00:00:00 +0000

Production resource limits

systemd owns resource limits and process lifecycle for huatuo-bamai.service. The service unit disables Huatuo self-managed cgroups and uses native controls:

[Service]
CPUAccounting=yes
CPUQuota=200%
MemoryAccounting=yes
MemoryMax=2G
TasksAccounting=yes
TasksMax=32768
KillMode=control-group

CPUQuota=200% allows up to 2 CPU cores and MemoryMax=2G caps service memory. Adjust these values for the host, collection jobs, and observed peaks. Do not pass --enable-cgroup in a systemd deployment; it would move the process out of the service cgroup.