Runner can't connect to broker.actions.githubusercontent.com

Describe the bug

Runner can't connect to broker.actions.githubusercontent.com and fail after 100 seconds

To Reproduce

Steps to reproduce the behavior:

1. Start the Runner using an AWS CodeBuild GitHub Runner
2. Run your workflows

Expected behavior

No random failures, no timeouts

Runner Version and Platform

Version of your runner?

runnerVersion=2.329.0
os=Linux
architecture=ARM64

What's not working?

Randomly jobs fail with the following message:

The self-hosted runner lost communication with the server. Verify the machine is running and has a healthy network connection. Anything in your workflow that terminates the runner process, starves it for CPU/Memory, or blocks its network access can cause this error.

Job Log Output

N/A

Runner and Worker's Diagnostic Logs

/codebuild/output/src3345867885/src/actions-runner/_diag/Runner_20251030-103632-utc.log:[2025-10-30 10:50:38Z INFO JobDispatcher] Successfully renew job 94f56804-2bdd-591d-882e-c21c9b8f5963, job is valid till 10/30/2025 11:00:38

/codebuild/output/src3345867885/src/actions-runner/_diag/Runner_20251030-103632-utc.log:[2025-10-30 10:51:54Z INFO JobDispatcher] Successfully renew job 94f56804-2bdd-591d-882e-c21c9b8f5963, job is valid till 10/30/2025 11:01:46

/codebuild/output/src3345867885/src/actions-runner/_diag/Runner_20251030-103632-utc.log:[2025-10-30 11:01:21Z WARN GitHubActionsService] Retrieving an AAD auth token took a long time (3.9861763 seconds)

/codebuild/output/src3345867885/src/actions-runner/_diag/Runner_20251030-103632-utc.log:[2025-10-30 11:23:37Z WARN GitHubActionsService] GET request to https://broker.actions.githubusercontent.com/message?sessionId=9964b969-253e-4c72-801a-769f117b4f0a&status=Busy&runnerVersion=2.329.0&os=Linux&architecture=ARM64&disableUpdate=true timed out after 100 seconds.

/codebuild/output/src3345867885/src/actions-runner/_diag/Runner_20251030-103632-utc.log:[2025-10-30 11:23:37Z WARN GitHubActionsService] POST request to https://run-actions-2-azure-eastus.actions.githubusercontent.com/66/renewjob timed out after 100 seconds.

Start by reproducing the failure with an AWS CodeBuild GitHub Runner, then inspect the diagnostic log at /codebuild/output/src3345867885/src/actions-runner/_diag/Runner_20251030-103632-utc.log around the broker and renewjob timeout entries. Done means identifying the cause of the intermittent 100-second connection failures and establishing a verified corrective change or documented root cause.