My Opensearch Indices mysteriously disappear

For a single node, T3.small, can you check the CloudWatch, your cluster metrics of "Nodes". This is the node count metrics for your cluster.

If you see the value fluctuates, it means that your node dropped. It later self-heals. It can be that you run out of CPU credits for burstable instance type of T type family.

To access the dashboard, OpenSearch keeps the user data in internal index, i.e. .opendistro-*. If you are using single node, and node drops, then the data will lose. Hence, all of your user data/user name will not have backup. Hence you cannot access the dashboard again.

To gain access to dashboard again, you will need to add master user again in the AWS Console.

It's best to not use T3 instance type. If it's just for test / development purpose, it's advised to have at least 3 data nodes. This way, you will at least have a replica on different nodes.