使用 ansible was_ec2 动态库存插件到达堡垒主机后面的盒子时出现问题

Question

我稍微绕了一下，我可以说这个 post 不是重复的。我一直在使用 Ansible 2.9.x 并且使用 ec2.py 动态清单连接到 bastion host 对我来说一直很好。我正在切换到 ansible was_ec2 插件，其中一个原因甚至是我的另一个 post。

我收集到的信息如下是我的库存文件和 ansible.cfg 文件

#myprovile.aws_ec2.yml
plugin: amazon.aws.aws_ec2
boto_profile: my profile
strict: True
regions:
  - eu-west-1
  - eu-central-1
  - eu-north-1
keyed_groups:
  - key: tags
    prefix: tag
hostnames:
  - ip-address
  # - dns-name
  # - tag:Name
  - private-ip-address
compose:
  ansible_host: private_ip_address

# folder/project level ansible.cfg  configuration
[defaults]
roles_path            = roles
host_key_checking     = False
hash_behaviour        = merge ### Note to self: Extremely important settings
interpreter_python    = auto  ### Note to self: Very important settings for running    from localhost

[inventory]
enable_plugins = aws_ec2, host_list, script, auto, yaml, ini, toml
# inventory = plugin_inventory/bb.aws_ec2.yaml

库存有 group_vars 个文件

➜  plugin_inventory git:(develop) ✗ tree
.
├── myprovile.aws_ec2.yml
└── group_vars
    ├── tag_Name_main_productname_uat_jumpbox.yml
    ├── tag_Name_main_productname_uat_mongo.yml
    ├── tag_Name_main_productname_uat_mongo_arb.yml
    ├── tag_Name_main_productname_uat_mysql.yml
    └── tag_Name_xxx.yml
    └── tag_Name_yyy.yml

要访问私有子网中的 mongo 数据库，group_vars 文件如下所示

#ansible_ssh_common_args: '-o ProxyCommand="ssh -o StrictHostKeyChecking=no -i {{ hostvars.localhost.reg_jumpbox_ssh_key }} -W %h:%p -q ubuntu@{{ hostvars.localhost.reg_jumpbox_facts.instances.0.public_ip_address }}"'
ansible_ssh_common_args: '-o ProxyCommand="ssh -o StrictHostKeyChecking=no -i ~/Dropbox/creds/pemfiles/ProductUATOps.pem -W %h:%p -q ubuntu@xxx.xxx.xxx.xxx"'

每次我运行命令

AWS_PROFILE=myprofile ansible -i ~/infrastructure_as_code/ansible_projects/productname/plugin_inventory/myprofile.aws_ec2.yml tag_Name_main_productname_uat_mongo -m ping -u ubuntu --private-key ~/Dropbox/creds/pemfiles/ProductUATOps.pem -vvvv

它没有连接，完整的输出和一些其他信息在 pastebin。

现在我看到的奇怪的是，即使在 ansible.cfg 中有 host_key_checking= False 我仍然在命令 Are you sure you want to continue connecting (yes/no/[fingerprint])?.

中找到以下内容

我也看到它正在寻找 ~/.ssh/known_hosts2\、/etc/ssh/ssh_known_hosts 和 /etc/ssh/ssh_known_hosts2，但那里有 ~/.ssh/known_hosts。

日志中还有一个令人困惑的错误 "module_stdout": "/bin/sh: 1: /Users/joseph/.pyenv/shims/python: not found\r\n"。但是python安装和pyenv已经一致了，oswise:

➜  ~ which python
/Users/joseph/.pyenv/shims/python

➜  ~ python --version
Python 3.8.12 (9ef55f6fc369, Oct 25 2021, 05:10:01)
[PyPy 7.3.7 with GCC Apple LLVM 13.0.0 (clang-1300.0.29.3)]

➜  ~ ls -lh /Users/joseph/.pyenv/shims/python
-rwxr-xr-x  1 joseph  staff   183B Feb 14 22:47   /Users/joseph/.pyenv/shims/python

➜  ~ /usr/bin/env python --version
Python 3.8.12 (9ef55f6fc369, Oct 25 2021, 05:10:01)
[PyPy 7.3.7 with GCC Apple LLVM 13.0.0 (clang-1300.0.29.3)]

我怀疑错误是由于某些东西阻止了指纹进入已知的 hosts 文件，我很想自己手动模拟 ssh 隧道，但我想了解为什么会这样，是否因为这是一台新机器。任何人都可以为我阐明这一点。谢谢

Answer 1

在运行 ansible-config dump 使用那个 ansible.cfg 之后，它发出 AnsibleOptionsError: Invalid value "merge ##... 所以它看起来 ansible 只是默默地吃掉了配置文件，或者可能正在使用不同的配置文件

虽然 # 是受支持的行首注释字符，但 ansible-config（自 2.12.1 起）仅允许 ; 作为行尾注释字符

[defaults]
roles_path            = roles
host_key_checking     = False
hash_behaviour        = merge ;;; Note to self: Extremely important settings
interpreter_python    = auto  ;;; Note to self: Very important settings for running    from localhost

[inventory]
enable_plugins = aws_ec2, host_list, script, auto, yaml, ini, toml

使用 ansible was_ec2 动态库存插件到达堡垒主机后面的盒子时出现问题

Issue reaching boxes behind a bastion host with ansible was_ec2 dynamic inventory plugin

ssh

amazon-ec2

ssh-tunnel

ansible

bastion-host