Help troubleshooting ECS Cluster update; appears to fail without error msg

I've been running a tast (webserver) on ECS for quite a while, through 585 revisions. Yesterday, I followed my usual procedure to update the cluster: a.) build a new container, b.) push to ECR, c.) create new revision in ECS console, d.) update cluster to new revision 586, e.) watch for new log stream to appear in CloudWatch to verify startup and check for errors. However, in this case no log stream appeared. Backing up to the ECS console, the task running the new revision had disappeared. I rebuilt and tested the container locally, re-pushed a new version to ECR, created a new task revision, and updated again, with the same result. The new task simply disappears from the Cluster->Tasks tab in the ECR console. No log stream is created for the task. The Cluster->Tasks->Logs tab only shows output from the last running task, Revision 585. The Cluster->Service->Events log shows the 'Start Task' event but no errors or terminations. No ECS Fargate instance is created, so nothing to exec into. How can I find out what's broken?