myxce/src/controller/testsuite/test_ha_repl_push.c

/*
 * test_ha_repl_push.c — Unity tests for Task 4.2:
 * Config and WAL replication push to standbys in the COMMITTING path.
 *
 * Tests verify:
 *   Req 7.1: When entering CTRL_PRIMARY_COMMITTING, push_config_to_standbys
 *            is called (via ha_push_config_hook seam).
 *   Req 7.2: When entering CTRL_PRIMARY_COMMITTING, ha_send_ctrlhello_to_all_standbys
 *            is called to trigger catch-up replication (via ha_send_ctrlhello_hook seam).
 *   Req 7.3: The push is triggered at the point of COMMITTING entry, not before.
 *            When election is lost (CTRL_STANDBY), no push occurs.
 *
 * Design ref: design.md "ha_send_ctrlhello_to_all_standbys" section
 * Boundary: ha.c (election evaluation + COMMITTING path)
 * Requirements: 7.1, 7.2, 7.3
 */
#define _GNU_SOURCE
#include <errno.h>
#include <semaphore.h>
#include <stdatomic.h>
#include <stdbool.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <time.h>

#include <unity/unity.h>

#include "ha.h"

/* ── Test-only seam: ha_send_ctrlhello_to_all_standbys call hook ─────────────
 *
 * In production, ha_elect_evaluate_and_signal calls
 * ha_send_ctrlhello_to_all_standbys(c) in the COMMITTING path.
 * For unit tests, the real function is not linked (cluster.c not in test binary).
 * The XCE_TESTING seam in ha.c checks ha_send_ctrlhello_hook; if non-NULL, it
 * calls the hook instead of the real ha_send_ctrlhello_to_all_standbys.
 */
extern int (*ha_send_ctrlhello_hook)(void *c);

/* ── Test-only seam: push_config_to_standbys call hook ──────────────────────
 *
 * In production, ha_elect_evaluate_and_signal calls push_config_to_standbys(c,...)
 * in the COMMITTING path.  For unit tests, the seam in ha.c checks
 * ha_push_config_hook; if non-NULL, it calls the hook instead.
 */
extern int (*ha_push_config_hook)(void *c);

/* ── Call counters for hooks ─────────────────────────────────────────────── */
static int g_ctrlhello_call_count = 0;
static int g_push_config_call_count = 0;

static int stub_ha_send_ctrlhello(void *c)
{
    (void)c;
    g_ctrlhello_call_count++;
    return 0;
}

static int stub_ha_push_config(void *c)
{
    (void)c;
    g_push_config_call_count++;
    return 0;
}

/* Also need the phase2 hook since ha_elect_evaluate_and_signal calls it */
extern int (*ha_phase2_hook)(void *c);

static int stub_phase2(void *c)
{
    (void)c;
    return 0;
}

/* ── Minimal ha_config_t for tests ─────────────────────────────────────── */

static ha_config_t g_cfg = {
    .heartbeat_interval_ms     = 3000,
    .heartbeat_timeout_ms      = 10000,
    .election_timeout_ms       = 2000,
    .probe_connect_timeout_ms  = 500,
    .max_promotion_retries     = 5,
    .election_jitter_max_ms    = 0,
    .election_priority_step_ms = 500,
    .committing_timeout_ms     = 10000,
};

/* ── Helper: build a minimal ha_elect_ctx_t ─────────────────────────────── */

static ha_elect_ctx_t *make_ctx(int n_other, int votes_granted)
{
    ha_elect_ctx_t *ctx = calloc(1, sizeof(*ctx) +
                                    (size_t)n_other * sizeof(ctx->votes[0]));
    if (!ctx) return NULL;
    ctx->n_peers       = n_other;
    ctx->votes_granted = votes_granted;
    ctx->n_replied     = n_other;
    ctx->max_resp_tag  = 0;
    ctx->max_resp_lsn  = 0;
    ctx->result        = -1;
    ctx->done_posted   = 0;
    ctx->c             = NULL; /* no real cluster in unit tests */
    pthread_mutex_init(&ctx->lock, NULL);
    sem_init(&ctx->done, 0, 0);
    if (n_other > 0) {
        ctx->votes[0].my_tag = 100;
        ctx->votes[0].my_lsn = 50;
    }
    return ctx;
}

static void free_ctx(ha_elect_ctx_t *ctx)
{
    if (!ctx) return;
    pthread_mutex_destroy(&ctx->lock);
    sem_destroy(&ctx->done);
    free(ctx);
}

/* ── Setup / teardown ────────────────────────────────────────────────────── */

static void reset_counters(void)
{
    g_ctrlhello_call_count  = 0;
    g_push_config_call_count = 0;
    ha_send_ctrlhello_hook  = stub_ha_send_ctrlhello;
    ha_push_config_hook     = stub_ha_push_config;
    ha_phase2_hook          = stub_phase2;
}

static void clear_hooks(void)
{
    ha_send_ctrlhello_hook  = NULL;
    ha_push_config_hook     = NULL;
    ha_phase2_hook          = NULL;
}

/* ── Test 1: Req 7.2 — COMMITTING path calls ha_send_ctrlhello_to_all_standbys ── */

static void test_committing_calls_ctrlhello_push(void)
{
    /* When election is won (CTRL_PRIMARY → COMMITTING), ha_send_ctrlhello_to_all_standbys
     * must be called exactly once.
     * Single-node wins: n_other=0, votes_granted=1. */
    reset_counters();

    ha_state_t ha;
    ha_init(&ha, &g_cfg, 0);
    ha.phase2_done = 1; /* skip phase2 */
    atomic_store_explicit(&ha.role, (int32_t)CTRL_ELECTING, memory_order_release);

    ha_elect_ctx_t *ctx = make_ctx(0, 1); /* single-node wins */
    TEST_ASSERT_NOT_NULL_MESSAGE(ctx, "make_ctx failed");

    ha_elect_evaluate_and_signal(&ha, ctx);

    TEST_ASSERT_EQUAL_INT_MESSAGE(1, g_ctrlhello_call_count,
        "ha_send_ctrlhello_to_all_standbys must be called once on COMMITTING entry (Req 7.2)");

    free_ctx(ctx);
    ha_fini(&ha);
    clear_hooks();
    printf("test_committing_calls_ctrlhello_push: PASS (ctrlhello_calls=%d)\n",
           g_ctrlhello_call_count);
}

/* ── Test 2: Req 7.1 — COMMITTING path calls push_config_to_standbys ──────── */

static void test_committing_calls_push_config(void)
{
    /* When election is won, push_config_to_standbys must be called once. */
    reset_counters();

    ha_state_t ha;
    ha_init(&ha, &g_cfg, 0);
    ha.phase2_done = 1;
    atomic_store_explicit(&ha.role, (int32_t)CTRL_ELECTING, memory_order_release);

    ha_elect_ctx_t *ctx = make_ctx(0, 1);
    TEST_ASSERT_NOT_NULL_MESSAGE(ctx, "make_ctx failed");

    ha_elect_evaluate_and_signal(&ha, ctx);

    TEST_ASSERT_EQUAL_INT_MESSAGE(1, g_push_config_call_count,
        "push_config_to_standbys must be called once on COMMITTING entry (Req 7.1)");

    free_ctx(ctx);
    ha_fini(&ha);
    clear_hooks();
    printf("test_committing_calls_push_config: PASS (push_config_calls=%d)\n",
           g_push_config_call_count);
}

/* ── Test 3: Req 7.3 — losing election → no push ───────────────────────── */

static void test_standby_result_does_not_push(void)
{
    /* When the election is lost (CTRL_STANDBY), neither push function is called. */
    reset_counters();

    ha_state_t ha;
    ha_init(&ha, &g_cfg, 0);
    ha.phase2_done = 1;
    atomic_store_explicit(&ha.role, (int32_t)CTRL_ELECTING, memory_order_release);

    /* n_other=2, quorum=2, votes_granted=1 → loses */
    ha_elect_ctx_t *ctx = make_ctx(2, 1);
    TEST_ASSERT_NOT_NULL_MESSAGE(ctx, "make_ctx failed");
    ctx->votes[0].my_tag = 100;
    ctx->votes[0].my_lsn = 50;

    ctrl_role_t result = ha_elect_evaluate_and_signal(&ha, ctx);
    TEST_ASSERT_EQUAL_INT_MESSAGE((int)CTRL_STANDBY, (int)result,
        "election with votes_granted=1 of n_other=2 must lose");

    TEST_ASSERT_EQUAL_INT_MESSAGE(0, g_ctrlhello_call_count,
        "ha_send_ctrlhello_to_all_standbys must NOT be called when standby (Req 7.3)");
    TEST_ASSERT_EQUAL_INT_MESSAGE(0, g_push_config_call_count,
        "push_config_to_standbys must NOT be called when standby (Req 7.3)");

    free_ctx(ctx);
    ha_fini(&ha);
    clear_hooks();
    printf("test_standby_result_does_not_push: PASS "
           "(ctrlhello=%d, push_config=%d)\n",
           g_ctrlhello_call_count, g_push_config_call_count);
}

/* ── Test 4: 3-node win also triggers both pushes ───────────────────────── */

static void test_3node_win_calls_both_pushes(void)
{
    /* n_other=2, quorum=2, votes_granted=3 → wins.
     * Both push hooks must be called exactly once. */
    reset_counters();

    ha_state_t ha;
    ha_init(&ha, &g_cfg, 0);
    ha.phase2_done = 1;
    atomic_store_explicit(&ha.role, (int32_t)CTRL_ELECTING, memory_order_release);

    ha_elect_ctx_t *ctx = make_ctx(2, 3); /* votes_granted=3 (self + 2 peers) */
    TEST_ASSERT_NOT_NULL_MESSAGE(ctx, "make_ctx failed");
    ctx->votes[0].my_tag = 100;
    ctx->votes[0].my_lsn = 50;

    ctrl_role_t result = ha_elect_evaluate_and_signal(&ha, ctx);
    TEST_ASSERT_EQUAL_INT_MESSAGE((int)CTRL_PRIMARY, (int)result,
        "3-node win must return CTRL_PRIMARY");

    ctrl_role_t actual_role =
        (ctrl_role_t)atomic_load_explicit(&ha.role, memory_order_acquire);
    TEST_ASSERT_EQUAL_INT_MESSAGE((int)CTRL_PRIMARY_COMMITTING, (int)actual_role,
        "ha->role must be CTRL_PRIMARY_COMMITTING after winning");

    TEST_ASSERT_EQUAL_INT_MESSAGE(1, g_ctrlhello_call_count,
        "ha_send_ctrlhello_to_all_standbys must be called once (3-node win)");
    TEST_ASSERT_EQUAL_INT_MESSAGE(1, g_push_config_call_count,
        "push_config_to_standbys must be called once (3-node win)");

    free_ctx(ctx);
    ha_fini(&ha);
    clear_hooks();
    printf("test_3node_win_calls_both_pushes: PASS "
           "(ctrlhello=%d, push_config=%d)\n",
           g_ctrlhello_call_count, g_push_config_call_count);
}

/* ── Test 5: done_posted guard prevents double push ─────────────────────── */

static void test_done_posted_prevents_double_push(void)
{
    /* Second call to ha_elect_evaluate_and_signal on same ctx is a no-op
     * (done_posted guard): neither push hook should fire a second time. */
    reset_counters();

    ha_state_t ha;
    ha_init(&ha, &g_cfg, 0);
    ha.phase2_done = 1;
    atomic_store_explicit(&ha.role, (int32_t)CTRL_ELECTING, memory_order_release);

    ha_elect_ctx_t *ctx = make_ctx(0, 1);
    TEST_ASSERT_NOT_NULL_MESSAGE(ctx, "make_ctx failed");

    ha_elect_evaluate_and_signal(&ha, ctx); /* first: triggers pushes (count=1) */
    ha_elect_evaluate_and_signal(&ha, ctx); /* second: done_posted guard → no-op */

    TEST_ASSERT_EQUAL_INT_MESSAGE(1, g_ctrlhello_call_count,
        "done_posted guard must prevent double ctrlhello push");
    TEST_ASSERT_EQUAL_INT_MESSAGE(1, g_push_config_call_count,
        "done_posted guard must prevent double config push");

    free_ctx(ctx);
    ha_fini(&ha);
    clear_hooks();
    printf("test_done_posted_prevents_double_push: PASS "
           "(ctrlhello=%d, push_config=%d)\n",
           g_ctrlhello_call_count, g_push_config_call_count);
}

/* ── Suite entry point ───────────────────────────────────────────────────── */

void test_ha_repl_push(void)
{
    RUN_TEST(test_committing_calls_ctrlhello_push);
    RUN_TEST(test_committing_calls_push_config);
    RUN_TEST(test_standby_result_does_not_push);
    RUN_TEST(test_3node_win_calls_both_pushes);
    RUN_TEST(test_done_posted_prevents_double_push);
}