DB statement sanitization causes memory leaks

[railsharipov]

Describe the bug

DB statement sanitization is enabled by default: https://opentelemetry.io/docs/zero-code/java/agent/instrumentation/#db-statement-sanitization

I was debugging memory leak for an application that dynamically builds large SQL query strings and frequently executes them using JDBC. I saw a lot of humongous live objects accumulating in heap space. After analyzing a heap dump I saw that the root cause is in a sqlToStatementInfoCache field in a io.opentelemetry.javaagent.shaded.instrumentation.api.incubator.semconv.db.SqlStatementSanitizer class.

Please let me know what additional information I can provide to help diagnose this issue.

Steps to reproduce

An application that dynamically builds large SQL query strings and frequently executes them using JDBC.

Expected behavior

Memory usage returns to baseline when application is idling.

Actual behavior

Baseline memory usage increases after application runs jobs involving a lot of large SQL query executions.

Javaagent or library instrumentation version

2.11.0

Environment

JDK: Amazon Coretto 17
OS: Alpine 3.18

Additional context

While investigating I ran application with different garbage collectors. Disabling DB statement sanitization resolved the issue.

[laurit]

How large are these sql statements? Multiple megabytes?

[railsharipov]

How large are these sql statements? Multiple megabytes?

Quite large, I see statements with sizes to up to 15MB