Ansible-inventory gcp_compute插件下划线添加到主机组名称中

Question

我正在尝试使用 gcp_compute 插件通过 GCE 标签解析一个 ansible 动态库存文件。我花了一段时间来设法运行插件，以便我实际上正确获取主机，但是由于某种原因，ansible-inventory 会在组名称中添加下划线前缀，即使前缀留空也是如此。

每当我使用以下 inventory.gcp.yml 文件运行 ansible-inventory 时：

plugin: gcp_compute
projects:
  - terzbx
filters:
keyed_groups:
  - key: tags['items']
auth_kind: serviceaccount
service_account_file: /root/deployment/MY_SERVICEACCOUNT.json

我收到的输出是：

 "all": {
     "children": [
         "__built_in_method_items_of_dict_object_at_0x2bc4db0_",
         "_analytics",
         "_consul_server",
         "_dashboardapi",
         "_elasticsearch",
         "_http_server",
         "_https_server",
         "_inventory",
         "_k2bq",
         "_kafka",
         "_kibana",
         "_logstash_k2bq",
         "_logstash_k2es",
         "_mysql",
         "_neo4j",
         "_ssl_offload",
         "_zookeeper",
         "ungrouped"
     ]

我希望它是：

 "all": {
     "children": [
         "_built_in_method_items_of_dict_object_at_0x2bc4db0_",
         "analytics",
         "consul_server",
         "dashboardapi",
         "elasticsearch",
         "http_server",
         "https_server",
         "inventory",
         "k2bq",
         "kafka",
         "kibana",
         "logstash_k2bq",
         "logstash_k2es",
         "mysql",
         "neo4j",
         "ssl_offload",
         "zookeeper",
         "ungrouped"
     ]

以前有人遇到过这个问题吗？我能做些什么来获得所需的输出吗？

提前感谢您对:)的帮助

Answer 1

偶然

遇到同样的问题，你应该试一试：

plugin: gcp_compute
projects:
  - terzbx
filters:
keyed_groups:
  - key: tags['items']
    separator: ''
auth_kind: serviceaccount
service_account_file: /root/deployment/MY_SERVICEACCOUNT.json