APM server logs contain document body from a partially failed bulk index request. For example, in case of unavailable_shards_exception for a specific document, since the ES response line contains the document body, and that APM server logs the ES response line on error, the document is effectively logged.

An issue was discovered whereby APM Server could log at ERROR level, a response from Elasticsearch indicating that indexing the document failed and that response would contain parts of the original document. Depending on the nature of the document that the APM Server attempted to ingest, this could lead to the insertion of sensitive or private information in the APM Server logs.

It was discovered that when acting as TLS clients, Beats, Elastic Agent, APM Server, and Fleet Server did not verify whether the server certificate is valid for the target IP address; however, certificate signature validation is still performed. More specifically, when the client is configured to connect to an IP address (instead of a hostname) it does not validate the server certificate's IP SAN values against that IP address and certificate validation fails, and therefore the connection is not blocked as expected.