Describe your experience troubleshooting JVM issues.
For this hands-on exercise, implement a detector that scans JVM metric samples and reports issue types that occur in at least two consecutive samples. Each sample contains heap_used, heap_max, gc_pause_ms, and thread_count.
Return [issue_name, start_index] pairs for the first qualifying run of each issue, ordered by index and then issue name. Use these thresholds: heap usage at least 90%, GC pause at least 500 ms, and thread count at least 1,000.
def detect_jvm_issues(samples):