Troubleshoot session recordings
Use this page to diagnose and resolve common Boundary session recording issues, including stale recordings after AWS credential rotation, unsupported RDP recovery workflows, and S3 storage access failures.
Known bugs
Rotation of AWS access and secret keys during a session results in stale recordings:
When you rotate a storage bucket's secrets, any new sessions use the new credentials. However, previously established sessions continue to use the old credentials.
As a best practice, administrators should rotate credentials in a phased manner, ensuring that all previously established sessions are completed before revoking the stale credentials. Otherwise, you may end up with recordings that aren't stored in the remote storage bucket and are unable to be played back.
Unsupported RDP recovery workflow during worker failure:
If a worker fails during an RDP recording, there is no way to recover the recording. This could happen due to a network connectivity issue or because a worker is scaled down, for example. There are a number of currently unsupported failure cases, including when the worker:
- Crashes and restarts while servicing connections and channels for session to an RDP target
- Crashes and restarts while closing a channel before all files are synced to remote storage
- Crashes and restarts while closing a connection recorder before all files are synced to remote storage
- Crashes and restarts while closing a session recorder before all files are synced to remote storage
- Crashes and never restarts or is seen again while closing a channel/connection recorder/session recorder
- Loses its connection to the controller and cancels sessions; it must resend the information when communication resumes
Interrupted SSH recordings may be able to be recovered. Refer to Session recording recovery for more information.
Debugging S3 bucket access issues
When you create or update a storage bucket, the actions GetBucketAcl and GetEncryptionConfiguration can be helpful in debugging access issues.