Tacker Helm 部署自动化¶

概述¶

通过 OpenStack-Helm 安装 Tacker 涉及多个手动步骤，包括系统先决条件的设置、环境配置以及各种组件的部署。这个过程容易出错且耗时，特别是对于新用户而言。

为了简化此过程，引入了一个基于 Python 的自动化脚本，通过处理所有先决条件和部署步骤，简化并优化了使用 OpenStack-Helm 安装 Tacker 的过程。

自动化脚本可在 Tacker 仓库中找到

tacker/tools/tacker_helm_deployment_automation_script

本文档提供了使用自动化脚本进行基于 Helm 的 Tacker 安装的步骤和指南。

先决条件¶

所有节点必须能够访问 OpenStack 仓库。仓库将在脚本执行期间自动下载。

$ git clone https://opendev.org/openstack/openstack-helm.git
$ git clone https://opendev.org/zuul/zuul-jobs.git

所有参与节点应相互连接。

配置更改¶

编辑 k8s_env/inventory.yaml 以进行 k8s 部署

添加以下内容以运行 ansible

# ansible user for running the playbook
ansible_user: <USERNAME>
ansible_ssh_private_key_file: <PATH_TO_SSH_KEY_FOR_USER>
ansible_ssh_extra_args: -o StrictHostKeyChecking=no

添加用于运行 kubectl 和 helm 命令的用户和组。

# The user and group that will be used to run Kubectl and Helm commands.
kubectl:
  user: <USERNAME>
  group: <USERGROUP>
# The user and group that will be used to run Docker commands.
docker_users:
  - <USERNAME>

添加将在主节点和工作节点之间用于通信的用户详细信息。必须将此用户配置为在所有节点上使用无密码 ssh。
```
# to connect to the k8s master node via ssh without a password.
client_ssh_user: <USERNAME>
cluster_ssh_user: <USER_GROUP>
```

如果使用裸机服务器进行负载均衡，请启用 MetalLB。

# MetalLB controllr is used for bare-metal loadbalancer.
metallb_setup: true

如果部署 CEPH 集群，请启用循环设备配置以供 CEPH 使用

# Loopback devices will be created on all cluster nodes which then can be used
# to deploy a Ceph cluster which requires block devices to be provided.
# Please use loopback devices only for testing purposes. They are not suitable
# for production due to performance reasons.
loopback_setup: true
loopback_device: /dev/loop100
loopback_image: /var/lib/openstack-helm/ceph-loop.img
loopback_image_size: 12G

添加将安装 Kubectl 和 Helm 的主节点。

children:
# The primary node where Kubectl and Helm will be installed. If it is
# the only node then it must be a member of the groups k8s_cluster and
# k8s_control_plane. If there are more nodes then the wireguard tunnel
# will be established between the primary node and the k8s_control_plane
  node.
primary:
  hosts:
    primary:
      ansible_host: <PRIMARY_NODE_IP>

添加将部署 Kubernetes 集群的节点。如果只有一个节点，则在此处提及它

# The nodes where the Kubernetes components will be installed.
k8s_cluster:
  hosts:
   primary:
     ansible_host: <IP_ADDRESS>
   node-2:
     ansible_host: <IP_ADDRESS>
   node-3:
     ansible_host: <IP_ADDRESS>

在 section 中添加控制平面节点

# The control plane node where the Kubernetes control plane components
  will be installed.
# It must be the only node in the group k8s_control_plane.
k8s_control_plane:
  hosts:
    primary:
      ansible_host: <IP_ADDRESS>

在 k8s_nodes section 中添加工作节点，如果是一节点安装，则保留该 section 为空

# These are Kubernetes worker nodes. There could be zero such nodes.
# In this case the Openstack workloads will be deployed on the control
  plane node.
k8s_nodes:
  hosts:
    node-2:
      ansible_host: <IP_ADDRESS>
    node-3:
      ansible_host: <IP_ADDRESS>

您可以在 [1] 中找到 inventory.yaml 的完整示例。

编辑 TACKER_NODE 在 config/config.yaml 中，使用主节点的 hostname 将其标记为控制平面节点
```
NODES:
 TACKER_NODE: <CONTROL-PLANE_NODE_HOSTNAME>
```

脚本执行¶

确保用户具有执行脚本的权限

$ ls -la Tacker_Install.py
 -rwxr-xr-x 1 root root 21923 Jul 22 10:00 Tacker_Install.py

执行命令以运行脚本
```
$ sudo python3 Tacker_install.py
```
执行命令以运行脚本

$ kubectl get pods -n openstack | grep -i  Tacker
  tacker-conductor-d7595d756-6k8wp            1/1     Running     0          24h
  tacker-db-init-mxwwf                        0/1     Completed   0          24h
  tacker-db-sync-4xnhx                        0/1     Completed   0          24h
  tacker-ks-endpoints-4nbqb                   0/3     Completed   0          24h
  tacker-ks-service-c8s2m                     0/1     Completed   0          24h
  tacker-ks-user-z2cq7                        0/1     Completed   0          24h
  tacker-rabbit-init-fxggv                    0/1     Completed   0          24h
  tacker-server-6f578bcf6c-z7z2c              1/1     Running     0          24h

有关详细信息，请参阅 [2] 中的文档。

Tacker Helm 部署自动化