ansible-lint 6.8.3+ does a fetch to api.github.com at the end of every run to check the latest version. This causes an excessive slowdown (120 seconds per role) if the system is dropping all outbound connections.

The symptom is slightly better on galaxy-importer 0.4.11 because we only call ansible-lint once instead of against every role individually, but the problem still remains. It's a hardcoded 120 second timeout and there's no configurable option to remedy it.

We should always pass --offline to ansible-lint calls from galaxy-importer unless there's a very good reason not to. Perhaps it can be a config option with a default false value.

The version check function ...
https://github.com/ansible/ansible-lint/commit/08a89892121d92b98b3f837116d5cea71b088bd4

where 0.4.10 makes the ansible-lint call ...
https://github.com/ansible/galaxy-importer/blob/abed2e9c0c918a9577496c26dd1650e105332137/galaxy_importer/loaders/content.py#L219-L229

where 0.4.11+ makes the ansible-lint call ...
https://github.com/ansible/galaxy-importer/blob/3e1d462c5969903be5121cc71ebc27a5eb0ed113/galaxy_importer/loaders/collection.py#L125-L137