If you discover a potential security issue in this project we ask that you notify OpenSearch Security directly via email to security@opensearch.org. Please do not create a public GitHub issue.
We strive to make these tools reliable and secure to use. However, before you begin using them with production or potentially sensitive data, we'd like to highlight a few points you should be aware of.
When configuring the Traffic Replayer, you can provide a set of credentials that will be attached to the replayed requests. It could be possible for an attacker to send a series of requests that they are not authorized to execute against the source cluster. The requests would be denied on the source cluster, but replayed against the target cluster with the configured credentials which may have higher privileges and allow the requests to be executed. The attacker could not directly read data this way, but any side effects of the requests (e.g. deleting data) would be executed.
In this case, an attacker could send a series of large and potentially malformed requests to the source cluster via the capture proxy. These messages would be relayed to the Kafka cluster, and if they were PUT/POST/UPDATE requests, block sending the request to the source cluster until the message was finished. If the attacker is able to use this strategy to tie up the proxy and/or Kafka cluster, all other incoming mutating requests to the source cluster would be blocked.
We have partially mitigated this by preventing the proxy from blocking for more than a fixed period of time (10 seconds by default, configurable in this file here, however the flow of messages to Kafka could still be disrupted.
If you are concerned about this scenario, we recommend fully mitigating it by putting a load-shedder in front of the proxy.
The output tuples, available on the shared EFS volume via the Migration Console, contain the exact requests and responses received from both the source and target clusters with the headers and the body of the messages. The Authorization header is present on SigV4 signed requests and those using basic authorization, and with basic authorization credentials can be extracted from the header value. These values are often essential for debugging and so are not censored from the output.
If you use basic authorization credentials, ensure that access to your output tuples is protected similarly to the credentials themselves.