AWS solutions documentation change
Summary
Added troubleshooting sections for live races and admin model downloads, updated CloudWatch log group references, and expanded operational diagnostics
Security assessment
The changes focus on operational troubleshooting (e.g., race evaluation failures, video streaming issues, leaderboard updates) and log group updates for diagnostics. There's no mention of vulnerabilities, security patches, access controls beyond existing roles, or data protection mechanisms. Changes like clarifying admin role requirements and rate limits are routine operational documentation.
Diff
diff --git a/solutions/latest/deepracer-on-aws/troubleshooting-the-solution.md b/solutions/latest/deepracer-on-aws/troubleshooting-the-solution.md index efcdb6512..19a36ec78 100644 --- a//solutions/latest/deepracer-on-aws/troubleshooting-the-solution.md +++ b//solutions/latest/deepracer-on-aws/troubleshooting-the-solution.md @@ -7 +7 @@ -Issues with logging in or creating an accountIssues with creating a modelIssues with viewing the model listIssues with training or evaluating a modelIssues with participating in a raceIssues with importing or exporting a modelUsing CloudWatch Logs to diagnose issuesUsing CloudWatch Logs Insights queries to diagnose issuesUsing point-in-time recovery to restore data from a backup +Issues with logging in or creating an accountIssues with creating a modelIssues with viewing the model listIssues with training or evaluating a modelIssues with participating in a raceIssues with live racesIssues with importing or exporting a modelIssues with admin model downloadsUsing CloudWatch Logs to diagnose issuesUsing CloudWatch Logs Insights queries to diagnose issuesUsing point-in-time recovery to restore data from a backup @@ -64 +64 @@ This means that an error occurred while attempting to train your model, contact -For cloud admins, try examining the `DeepRacerWorkflow` log group that is associated with your deployment in the Amazon CloudWatch Logs console. +For cloud admins, try examining the `/aws/lambda/DeepRacerTraining` log group that is associated with your deployment in the Amazon CloudWatch Logs console. @@ -83,0 +84,35 @@ This means that your model has not met the qualification requirements for the ra +## Issues with live races + + 1. _A model evaluation is showing a**Failed** status in the queue. What should I do?_ + +Click **Reset** on the failed queue item. The system will take approximately 30 seconds to clean up the previous evaluation environment, after which the item returns to **Pending** status. You can then click **Launch** (or re-enable Autolaunch) to retry the evaluation. + +If the same model fails again after a reset, click **Remove** to remove it from the queue and proceed with the remaining submissions. For repeated or unexpected failures, examine the `/aws/lambda/DeepRacerTraining` and `/aws/lambda/DeepRacerLiveRacing` log groups in Amazon CloudWatch Logs. See Using CloudWatch Logs to diagnose issues. + + 2. _The live race video stream is not loading or is stuck buffering_ + +The video stream is delivered via Amazon Kinesis Video Streams as an HLS stream. Try the following steps: + + * Refresh the page to re-establish the stream connection. + + * Verify that the participant’s browser supports HLS playback (modern Chromium-based browsers and Safari are recommended). + + * If the stream does not recover, the facilitator can click **Reset** on the currently in-progress queue item to restart the evaluation, which will re-establish the stream. + +For cloud admins, check the `/aws/lambda/DeepRacerLiveRacing` log group for errors related to Kinesis Video Streams. + + 3. _The leaderboard is not updating in real-time_ + +Real-time leaderboard updates are delivered over a WebSocket connection via AWS IoT Core. If the leaderboard stops updating: + + * Refresh the page — the interface will automatically re-fetch the latest race state on reconnect. + + * If the issue persists for multiple participants, a cloud admin should check whether IoT Core is experiencing connectivity issues in the deployed region. + + 4. _The**Launch** button is disabled and I cannot start the next evaluation_ + +The **Launch** button is disabled when: an evaluation is already in progress, the race has been completed (winner declared), or the current time is before the scheduled live event time. If the button is enabled but launching fails with an error, verify that the "Submissions Open" toggle is off and that there are pending items in the queue. If the race was not intentionally completed, contact your cloud admin to check the `/aws/lambda/DeepRacerLiveRacing` log group for unexpected state transitions. + + + + @@ -92,0 +128,17 @@ If you are using a Windows computer, you’ll need to download and install a fil +## Issues with admin model downloads + + 1. _I can’t see the Model Management page in the sidebar_ + +The Model Management page is only visible to users with the Admin or Race Facilitator role. Ask your instance administrator to update your role if you need access to this feature. + + 2. _I’m getting an "Access Denied" error when trying to download a model_ + +Verify that your account has the Admin or Race Facilitator role. If your role was recently changed, log out and log back in for the change to take effect. + + 3. _My download requests are being rejected with a rate limit error_ + +The admin model management endpoints are protected by a rate limit of 100 requests per 5 minutes per user. If you are downloading many models at once, wait a few minutes before retrying. + + + + @@ -123 +175,9 @@ This log group contains the log output from all functions that handle user authe -This log group contains the log output from the functions that are responsible for dispatching, initializing, monitoring, and finalizing training and evaluation jobs. It can be useful for diagnosing any errors that are thrown while a model is being trained or evaluated, or going through the various state changes expected as part of this process. +This log group contains the log output from the functions that handle model import workflows (dispatching, validating, and completing import operations). It can be useful for diagnosing issues with importing models. + +**/aws/lambda/DeepRacerTraining** + +This log group contains the log output from the functions that are responsible for dispatching, initializing, monitoring, and finalizing training and evaluation jobs. It can be useful for diagnosing any errors that are thrown while a model is being trained or evaluated, or during a community race submission evaluation. + +**/aws/lambda/DeepRacerLiveRacing** + +This log group contains the log output from all functions that support live race orchestration, including queue management, autolaunch, stream handling, and event state transitions. It is the primary log group for diagnosing issues with live races.