运行 Ansible playbook 时在 Ec2.py 中刷新 ElastiCache 时出错

Question

在过去的几天里，我在使用 AWS 云模块运行 ansible playbook 或从本地计算机使用 --refresh-cache 命令直接运行 ec2.py 文件时遇到以下错误：

ap:~ ap$ /ansible/inventory/ec2.py --refresh-cache
Traceback (most recent call last):
  File "/ansible/inventory/ec2.py", line 1510, in <module>
    Ec2Inventory()
  File "/ansible/inventory/ec2.py", line 186, in __init__
    self.do_api_calls_update_cache()
  File "/ansible/inventory/ec2.py", line 492, in do_api_calls_update_cache
    self.get_elasticache_replication_groups_by_region(region)
  File "/ansible/inventory/ec2.py", line 722, in get_elasticache_replication_groups_by_region
    self.add_elasticache_replication_group(replication_group, region)
  File "/ansible/inventory/ec2.py", line 1198, in add_elasticache_replication_group
    dest = replication_group['NodeGroups'][0]['PrimaryEndpoint']['Address']
TypeError: 'NoneType' object has no attribute '__getitem__'

从 AWS 账户中的 Ec2 实例运行时，此问题似乎消失了。

我有最新的 EC2.py 和 EC2.ini 文件，Ansible 版本是 2.2.0.0，安装了 boto3，并且我在运行 playbook 之前设置了有效的 AWS 凭证。

有什么想法吗？

Answer 1

在https://github.com/ansible/ansible/pull/20190尝试补丁

 @@ -1194,13 +1194,14 @@ def add_elasticache_replication_group(self, replication_group, region):
          if not self.all_elasticache_replication_groups and replication_group['Status'] != 'available':
              return

 +        # Skip clusters we cannot address (e.g. private VPC subnet or clustered redis)
 +        if replication_group['NodeGroups'][0]['PrimaryEndpoint'] is None or \
 +           replication_group['NodeGroups'][0]['PrimaryEndpoint']['Address'] is None:
 +            return
 +
          # Select the best destination address (PrimaryEndpoint)
          dest = replication_group['NodeGroups'][0]['PrimaryEndpoint']['Address']

 -        if not dest:
 -            # Skip clusters we cannot address (e.g. private VPC subnet)
 -            return
 -
          # Add to index
          self.index[dest] = [region, replication_group['ReplicationGroupId']]