java-topology/defects/clickhouse-java/unit/ClickHouseJavaTest.java

import java.util.*;

/**
 * CWE-407 simulation: ClickHouseNodes load-balancing faulty-node contains scan
 *
 * Defect: ClickHouseLoadBalancingPolicy.FirstAlivePolicy.get() iterates
 * manager.nodes (LinkedList<N>) and calls manager.faultyNodes.contains(node)
 * for every iteration — O(N×F) where N = healthy nodes, F = faulty nodes.
 *
 * Fix: change nodes and faultyNodes to LinkedHashSet<N> for O(1) contains.
 */
public class ClickHouseJavaTest {

    // --- Simulated defect: LinkedList for both collections ---
    static int simulateDefect(int healthyCount, int faultyCount) {
        LinkedList<String> nodes = new LinkedList<>();
        LinkedList<String> faultyNodes = new LinkedList<>();

        for (int i = 0; i < healthyCount; i++) nodes.add("node-" + i);
        for (int i = 0; i < faultyCount; i++) faultyNodes.add("faulty-" + i);

        int ops = 0;
        // Simulate FirstAlivePolicy.get() — finds first healthy non-faulty node
        for (String n : nodes) {
            ops++;
            if (!faultyNodes.contains(n)) {  // O(F) per iteration
                break;
            }
        }
        return ops;
    }

    // --- Simulated fix: LinkedHashSet for O(1) contains ---
    static int simulateFix(int healthyCount, int faultyCount) {
        LinkedHashSet<String> nodes = new LinkedHashSet<>();
        LinkedHashSet<String> faultyNodes = new LinkedHashSet<>();

        for (int i = 0; i < healthyCount; i++) nodes.add("node-" + i);
        for (int i = 0; i < faultyCount; i++) faultyNodes.add("faulty-" + i);

        int ops = 0;
        for (String n : nodes) {
            ops++;
            if (!faultyNodes.contains(n)) {  // O(1) via HashSet
                break;
            }
        }
        return ops;
    }

    // --- Measure wall-clock ops for contains cost comparison ---
    static long timeDefect(int healthyCount, int faultyCount, int iterations) {
        LinkedList<String> nodes = new LinkedList<>();
        LinkedList<String> faultyNodes = new LinkedList<>();

        // Make all healthy nodes also appear in faulty (worst case: always scan all)
        for (int i = 0; i < healthyCount; i++) nodes.add("node-" + i);
        // faultyNodes doesn't contain healthy nodes, but is large enough to cost
        for (int i = 0; i < faultyCount; i++) faultyNodes.add("faulty-" + i);

        long start = System.nanoTime();
        for (int iter = 0; iter < iterations; iter++) {
            for (String n : nodes) {
                if (!faultyNodes.contains(n)) break;
            }
        }
        return System.nanoTime() - start;
    }

    static long timeFix(int healthyCount, int faultyCount, int iterations) {
        LinkedHashSet<String> nodes = new LinkedHashSet<>();
        LinkedHashSet<String> faultyNodes = new LinkedHashSet<>();

        for (int i = 0; i < healthyCount; i++) nodes.add("node-" + i);
        for (int i = 0; i < faultyCount; i++) faultyNodes.add("faulty-" + i);

        long start = System.nanoTime();
        for (int iter = 0; iter < iterations; iter++) {
            for (String n : nodes) {
                if (!faultyNodes.contains(n)) break;
            }
        }
        return System.nanoTime() - start;
    }

    public static void main(String[] args) {
        System.out.println("=== CWE-407: ClickHouseNodes load-balancing faulty-node O(N×F) ===");
        System.out.println();

        // --- Correctness: ops to find first healthy node are the same ---
        int HEALTHY = 10, FAULTY = 500;
        int defectOps = simulateDefect(HEALTHY, FAULTY);
        int fixOps    = simulateFix(HEALTHY, FAULTY);
        System.out.printf("PASS: defect ops=%d fix ops=%d (same result)%n", defectOps, fixOps);
        assert defectOps == fixOps : "ops should match";

        // --- Performance: warm up then measure ---
        int ITER = 50_000;
        // Warm up
        for (int i = 0; i < 3; i++) { timeDefect(HEALTHY, FAULTY, ITER); timeFix(HEALTHY, FAULTY, ITER); }

        long defectNs = timeDefect(HEALTHY, FAULTY, ITER);
        long fixNs    = timeFix(HEALTHY, FAULTY, ITER);
        double ratio  = (double) defectNs / fixNs;

        System.out.printf("Defect (LinkedList.contains): %,d ns over %,d iterations%n", defectNs, ITER);
        System.out.printf("Fix    (HashSet.contains):    %,d ns over %,d iterations%n", fixNs, ITER);
        System.out.printf("Speedup ratio: %.1fx%n", ratio);

        // --- Complexity demonstration at N=10, F varies ---
        System.out.println();
        System.out.println("Complexity demo (N=10 healthy, varies F faulty):");
        System.out.printf("  %-6s  %-12s  %-12s  %-8s%n", "F", "Defect(ns)", "Fix(ns)", "Ratio");
        for (int f : new int[]{10, 50, 100, 250, 500, 1000}) {
            long d = timeDefect(10, f, ITER);
            long fx = timeFix(10, f, ITER);
            System.out.printf("  %-6d  %-12d  %-12d  %-8.1f%n", f, d, fx, (double) d / fx);
        }

        // --- Assert speedup is meaningful (at least 2x at F=500) ---
        assert ratio >= 2.0 : String.format("Expected speedup >= 2x, got %.1fx", ratio);
        System.out.println();
        System.out.println("PASS: all assertions satisfied");
    }
}