myxce/src/controller/testsuite/test_worker_join.c

/*
 * test_worker_join.c — TDD tests for worker join refactor
 *
 * Task 1.1: Failing tests for apply_join_state (shared status/resource update)
 * Task 1.2: Failing tests for init_worker (first-time join path)
 * Task 1.3: Failing tests for rejoin_worker (reconnect update with admin-bit preservation)
 *
 * RED phase: apply_join_state / init_worker / rejoin_worker are stubs that
 * return 0 without modifying state, so all assertions below fail until
 * Tasks 2–3 provide real implementations.
 */

/* rtti.h must come first — it defines 'var' and 'typeinfo_t' used by kvmgr.h */
#include <rtti.h>

#include <errno.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <unistd.h>

#include <unity/unity.h>

#include "cluster.h"
#include "kvmgr.h"
#include "resobj.h"
#include "wobj.h"
#include "worker.h"
#include "xce/cluster_def.h"
#include "xce/types.h"
#include "xstring.h"
#include "zarray.h"

/* ── Stubs for controller functions not under test ───────────────────── */

/*
 * collect_confvars is called by on_worker_join → worker_push_config.
 * Stubbed here so worker.c links without pulling in the full dataex.c.
 */
void collect_confvars(cluster_t* c, collect_check_fn_t fn, xlist_t* l)
{
    (void)c;
    (void)fn;
    (void)l;
}

/* ── Forward declarations for functions under test ──────────────────── */

/*
 * apply_join_state is non-static in worker.c when XCE_TESTING is defined.
 *
 * RED: the current stub body does nothing and returns 0, so assertions
 * that check the modified state will fail.
 */
int apply_join_state(cluster_t* c, worker_t* w, endpoint_t* ep,
                     xce_hostconfig_t* hconfig);

/*
 * init_worker and rejoin_worker are exported from worker.c.
 *
 * RED: stubs return 0 without registering or modifying any entry.
 */
int init_worker(cluster_t* c, endpoint_t* ep, const char* host,
                hostattrs_t* attrs, xce_hostconfig_t* hconfig);
int rejoin_worker(cluster_t* c, worker_t* w, endpoint_t* ep,
                  hostattrs_t* attrs, xce_hostconfig_t* hconfig);

/* ── Test fixture ────────────────────────────────────────────────────── */

/* Resource names and value types matching add_hostattrs_resvals in worker.c */
#define N_RES 9
static const char* g_res_names[N_RES] = {
    "type", "model", "cpuf", "maxcpus", "maxmem",
    "maxswap", "maxtmp", "ndisks", "slots"
};
static const int g_res_valtypes[N_RES] = {
    XCE_VAL_STRING, XCE_VAL_STRING,
    XCE_VAL_NUMERIC, XCE_VAL_NUMERIC, XCE_VAL_NUMERIC,
    XCE_VAL_NUMERIC, XCE_VAL_NUMERIC, XCE_VAL_NUMERIC,
    XCE_VAL_NUMERIC
};

typedef struct {
    cluster_t c;
    var       workermgr;
    var       resmgr;
} test_cluster_t;

static test_cluster_t g_tc;

static void setup_test_cluster(void)
{
    kvmgr_params_t str_params = {
        .key_type   = KVMGR_KEY_STRING,
        .nslots     = 64,
        .delta      = 16,
        .datakiller = NULL,
    };

    memset(&g_tc, 0, sizeof(g_tc));

    g_tc.workermgr = makeInstance(&xkvmgr_type, &str_params);
    TEST_ASSERT_NOT_NULL(g_tc.workermgr);

    g_tc.resmgr = makeInstance(&xkvmgr_type, &str_params);
    TEST_ASSERT_NOT_NULL(g_tc.resmgr);

    for (int i = 0; i < N_RES; i++) {
        xce_resource_t rdef;
        memset(&rdef, 0, sizeof(rdef));
        rdef.name    = (char*)g_res_names[i];
        rdef.valtype = g_res_valtypes[i];
        rdef.flags   = 0;
        rdef.idx     = -1;

        xce_resource_t* r = makeResource(g_tc.resmgr, &rdef);
        TEST_ASSERT_NOT_NULL_MESSAGE(r, g_res_names[i]);
    }

    g_tc.c.workermgr = g_tc.workermgr;
    g_tc.c.resmgr    = g_tc.resmgr;
    zarray_init(&g_tc.c.sharedinst, 64, NULL);
}

static void teardown_test_cluster(void)
{
    zarray_fini(&g_tc.c.sharedinst);
    if (g_tc.resmgr)    dereference(g_tc.resmgr);
    if (g_tc.workermgr) dereference(g_tc.workermgr);
    memset(&g_tc, 0, sizeof(g_tc));
}

/* Create a worker in the test workermgr; attrs strings must be makeString'd. */
static worker_t* make_test_worker(const char* name, hostattrs_t* attrs)
{
    xce_worker_t host;
    memset(&host, 0, sizeof(host));
    host.name  = (char*)name;
    host.attrs = attrs;
    return makeWorker(g_tc.workermgr, &host);
}

/* Retrieve a numeric resource value from a worker's resvals by resource name */
static float get_worker_res_fval(worker_t* w, const char* name)
{
    xce_resource_t* r = NULL;
    lock(g_tc.c.resmgr, RTTI_LOCK_READ);
    key_find(g_tc.c.resmgr, (void*)name, (void**)&r);
    unlock(g_tc.c.resmgr);
    if (r == NULL || r->idx < 0 || r->idx >= w->host.resvals.caps)
        return -1.0f;
    xce_val_t* v = (xce_val_t*)w->host.resvals.data[r->idx];
    return (v != NULL) ? v->u.fvalue : -1.0f;
}

/*
 * Capture whatever update_resvals()/update_resval_list() print during a
 * failing update_resval() call, as raw runtime output rather than a
 * source-text scan. This test binary never calls mylog_init()/tlog_init()
 * (only the real daemon main()s do -- see test_apply_correctness_
 * verification.c's own header comment, and the identical established
 * precedent in scheduler/testsuite/test_scheduler_connect.c's
 * capture_stdout_during_sched_init()), so tlog's internal root log handle
 * stays NULL and every ERROR()/WARN() call falls through tlog's early-
 * print path, which writes the fully formatted line directly to
 * STDOUT_FILENO via a raw write() -- genuine runtime log capture, not a
 * guess about which stream PERROR/ERROR happens to use.
 */
static char* capture_stdout_during(void (*fn)(void* ctx), void* ctx)
{
    char tmp[] = "/tmp/xce_worker_join_logcap_XXXXXX";
    int  fd    = mkstemp(tmp);
    TEST_ASSERT_TRUE_MESSAGE(fd >= 0,
                             "fixture: mkstemp for stdout capture must succeed");

    fflush(stdout);
    int saved_stdout = dup(STDOUT_FILENO);
    TEST_ASSERT_TRUE_MESSAGE(saved_stdout >= 0,
                             "fixture: dup(STDOUT_FILENO) must succeed");
    TEST_ASSERT_TRUE_MESSAGE(dup2(fd, STDOUT_FILENO) >= 0,
                             "fixture: dup2 redirect of STDOUT_FILENO must "
                             "succeed");

    fn(ctx);

    fflush(stdout);
    dup2(saved_stdout, STDOUT_FILENO);
    close(saved_stdout);

    off_t len = lseek(fd, 0, SEEK_CUR);
    TEST_ASSERT_TRUE_MESSAGE(len >= 0,
                             "fixture: lseek on capture file must succeed");

    char* buf = (char*)calloc(1, (size_t)len + 1);
    TEST_ASSERT_NOT_NULL(buf);
    if (len > 0) {
        TEST_ASSERT_TRUE_MESSAGE(lseek(fd, 0, SEEK_SET) == 0,
                                 "fixture: rewind of capture file must "
                                 "succeed");
        ssize_t nread = read(fd, buf, (size_t)len);
        TEST_ASSERT_TRUE_MESSAGE(nread == len,
                                 "fixture: reading back captured buffer "
                                 "must succeed");
    }
    close(fd);
    unlink(tmp);

    return buf;
}

typedef struct {
    cluster_t*    c;
    worker_t*     w;
    xce_resval_t* val;
} update_resvals_call_ctx_t;

static void call_update_resvals_1(void* ctx_)
{
    update_resvals_call_ctx_t* ctx = (update_resvals_call_ctx_t*)ctx_;
    update_resvals(ctx->c, ctx->w, 1, ctx->val);
}

/*
 * update_resval() returns its failure reason as a raw negative code
 * directly (key_find/zarray_put convention -- never via errno), so
 * update_resvals()/update_resval_list() logging it via PERROR (which reads
 * the process-global errno) reported whatever errno happened to be left
 * over from an unrelated earlier call: "Success" if nothing had touched
 * it, or a stray EAGAIN/etc from some other syscall in the same event-loop
 * iteration -- never the real reason a resource update failed. This test
 * reproduces that exact symptom (ambient errno=EAGAIN, unrelated to the
 * genuine failure) and proves the fix decodes the actual rc instead.
 */
static void test_update_resvals_logs_actual_rc_not_stale_errno(void)
{
    setup_test_cluster();

    hostattrs_t attrs;
    memset(&attrs, 0, sizeof(attrs));
    attrs.hosttype  = makeString("LINUX");
    attrs.hostmodel = makeString("x86");
    worker_t* w = make_test_worker("host-resval-log", &attrs);
    TEST_ASSERT_NOT_NULL(w);

    xce_val_t fval;
    memset(&fval, 0, sizeof(fval));
    fval.which    = Value_u_fvalue;
    fval.u.fvalue = 1.0f;

    xce_resval_t val;
    val.name = "does-not-exist-in-resmgr";
    val.val  = &fval;

    /* Simulate the exact reported symptom: ambient errno unrelated to the
     * real failure, left over from some other unrelated call. */
    errno = EAGAIN;

    update_resvals_call_ctx_t ctx = { .c = &g_tc.c, .w = w, .val = &val };
    char* captured = capture_stdout_during(call_update_resvals_1, &ctx);

    TEST_ASSERT_NOT_NULL(captured);
    TEST_ASSERT_NULL_MESSAGE(
        strstr(captured, "Resource temporarily unavailable"),
        "must not log the ambient errno's string (EAGAIN) -- that is not "
        "why update_resval failed");
    TEST_ASSERT_NULL_MESSAGE(
        strstr(captured, "Success"),
        "must not log \"Success\" for a call that returned rc<0");
    TEST_ASSERT_NOT_NULL_MESSAGE(
        strstr(captured, "does-not-exist-in-resmgr"),
        "must name which resource failed");
    TEST_ASSERT_NOT_NULL_MESSAGE(
        strstr(captured, "No such file or directory"),
        "must log the real reason (-ENOENT, from key_find on an "
        "unregistered resource name)");

    free(captured);
    teardown_test_cluster();
    printf("test_update_resvals_logs_actual_rc_not_stale_errno: PASS\n");
}

/* ── Task 1.1: Tests for apply_join_state ──────────────────────────── */

/*
 * After apply_join_state, XCE_WORKER_UNAVAIL must be cleared.  Req 4.1
 */
static void test_apply_join_state_clears_unavail(void)
{
    setup_test_cluster();

    hostattrs_t attrs;
    memset(&attrs, 0, sizeof(attrs));
    attrs.hosttype  = makeString("LINUX");
    attrs.hostmodel = makeString("x86");
    attrs.cpufactor = 1.0f;
    attrs.maxcpus   = 4;
    attrs.maxmem    = 16;
    attrs.maxswap   = 32;
    attrs.maxtmp    = 64;
    attrs.ndisks    = 1;

    worker_t* w = make_test_worker("host-ajs-1", &attrs);
    TEST_ASSERT_NOT_NULL(w);

    lock(w, RTTI_LOCK_WRITE);
    w->host.status |= XCE_WORKER_UNAVAIL;
    unlock(w);

    xce_hostconfig_t hconfig;
    memset(&hconfig, 0, sizeof(hconfig));
    hconfig.hosttype  = (char*)"!"; /* no type override */
    hconfig.hostmodel = (char*)"!"; /* no model override */
    hconfig.mxj       = 2;

    endpoint_t ep;
    memset(&ep, 0, sizeof(ep));

    lock(w, RTTI_LOCK_WRITE);
    int rc = apply_join_state(&g_tc.c, w, &ep, &hconfig);
    unlock(w);

    TEST_ASSERT_EQUAL_INT(0, rc);
    TEST_ASSERT_FALSE_MESSAGE(w->host.status & XCE_WORKER_UNAVAIL,
                              "UNAVAIL must be cleared after apply_join_state");

    teardown_test_cluster();
    printf("test_apply_join_state_clears_unavail: PASS\n");
}

/*
 * When hconfig provides non-"!" overrides, hosttype and hostmodel must
 * match those values after apply_join_state.  Req 4.2
 */
static void test_apply_join_state_applies_hosttype_override(void)
{
    setup_test_cluster();

    hostattrs_t attrs;
    memset(&attrs, 0, sizeof(attrs));
    attrs.hosttype  = makeString("OLD_TYPE");
    attrs.hostmodel = makeString("OLD_MODEL");
    attrs.cpufactor = 1.0f;
    attrs.maxcpus   = 2;
    attrs.maxmem    = 8;
    attrs.maxswap   = 16;
    attrs.maxtmp    = 32;
    attrs.ndisks    = 1;

    worker_t* w = make_test_worker("host-ajs-2", &attrs);
    TEST_ASSERT_NOT_NULL(w);

    xce_hostconfig_t hconfig;
    memset(&hconfig, 0, sizeof(hconfig));
    hconfig.hosttype  = (char*)"LINUX64";
    hconfig.hostmodel = (char*)"x86_64";
    hconfig.mxj       = 4;

    endpoint_t ep;
    memset(&ep, 0, sizeof(ep));

    lock(w, RTTI_LOCK_WRITE);
    int rc = apply_join_state(&g_tc.c, w, &ep, &hconfig);
    unlock(w);

    TEST_ASSERT_EQUAL_INT(0, rc);
    TEST_ASSERT_EQUAL_STRING_MESSAGE("LINUX64", w->host.attrs->hosttype,
                                     "hosttype must be overridden from hconfig");
    TEST_ASSERT_EQUAL_STRING_MESSAGE("x86_64", w->host.attrs->hostmodel,
                                     "hostmodel must be overridden from hconfig");

    teardown_test_cluster();
    printf("test_apply_join_state_applies_hosttype_override: PASS\n");
}

/*
 * After apply_join_state, 'slots' equals hconfig.mxj and numeric host-attr
 * resources match the worker's attrs.  Req 4.2, 4.3
 */
static void test_apply_join_state_updates_resource_values(void)
{
    setup_test_cluster();

    hostattrs_t attrs;
    memset(&attrs, 0, sizeof(attrs));
    attrs.hosttype  = makeString("LINUX");
    attrs.hostmodel = makeString("x86");
    attrs.cpufactor = 3.0f;
    attrs.maxcpus   = 8;
    attrs.maxmem    = 64;
    attrs.maxswap   = 128;
    attrs.maxtmp    = 256;
    attrs.ndisks    = 4;

    worker_t* w = make_test_worker("host-ajs-3", &attrs);
    TEST_ASSERT_NOT_NULL(w);

    xce_hostconfig_t hconfig;
    memset(&hconfig, 0, sizeof(hconfig));
    hconfig.hosttype  = (char*)"!";
    hconfig.hostmodel = (char*)"!";
    hconfig.mxj       = 6;

    endpoint_t ep;
    memset(&ep, 0, sizeof(ep));

    lock(w, RTTI_LOCK_WRITE);
    int rc = apply_join_state(&g_tc.c, w, &ep, &hconfig);
    unlock(w);

    TEST_ASSERT_EQUAL_INT(0, rc);

    TEST_ASSERT_EQUAL_FLOAT_MESSAGE(6.0f, get_worker_res_fval(w, "slots"),
                                    "slots must equal hconfig.mxj");
    TEST_ASSERT_EQUAL_FLOAT(3.0f, get_worker_res_fval(w, "cpuf"));
    TEST_ASSERT_EQUAL_FLOAT(8.0f, get_worker_res_fval(w, "maxcpus"));
    TEST_ASSERT_EQUAL_FLOAT(64.0f, get_worker_res_fval(w, "maxmem"));
    TEST_ASSERT_EQUAL_FLOAT(128.0f, get_worker_res_fval(w, "maxswap"));
    TEST_ASSERT_EQUAL_FLOAT(256.0f, get_worker_res_fval(w, "maxtmp"));
    TEST_ASSERT_EQUAL_FLOAT(4.0f, get_worker_res_fval(w, "ndisks"));

    teardown_test_cluster();
    printf("test_apply_join_state_updates_resource_values: PASS\n");
}

/*
 * hconfig.mxj == -1 is the schema's documented "auto-detect" sentinel
 * (hosts.schema.json: "max_slot": {"minimum": -1, "description": "-1 means
 * auto-detect"}) -- apply_join_state must fall back to the worker's own
 * reported maxcpus, NOT persist -1 (or, before the xceadmin mapping fix,
 * silently ship 0 and permanently disable scheduling on that host: mxj is
 * UInt32 on the wire, so an unset/absent max_slot used to decode as 0,
 * which is not < 0 and so never reached this fallback -- see
 * xcecommon.mapping.HOSTCONFIG_MXJ_AUTO_DETECT). This is the ONLY test in
 * this file that exercises the nslots<0 branch; every sibling test above
 * uses a positive hconfig.mxj and so cannot catch a regression here.
 */
static void test_apply_join_state_mxj_negative_one_falls_back_to_maxcpus(void)
{
    setup_test_cluster();

    hostattrs_t attrs;
    memset(&attrs, 0, sizeof(attrs));
    attrs.hosttype  = makeString("LINUX");
    attrs.hostmodel = makeString("x86");
    attrs.cpufactor = 1.0f;
    attrs.maxcpus   = 12;
    attrs.maxmem    = 64;
    attrs.maxswap   = 128;
    attrs.maxtmp    = 256;
    attrs.ndisks    = 2;

    worker_t* w = make_test_worker("host-ajs-autodetect", &attrs);
    TEST_ASSERT_NOT_NULL(w);

    xce_hostconfig_t hconfig;
    memset(&hconfig, 0, sizeof(hconfig));
    hconfig.hosttype  = (char*)"!";
    hconfig.hostmodel = (char*)"!";
    hconfig.mxj       = -1;

    endpoint_t ep;
    memset(&ep, 0, sizeof(ep));

    lock(w, RTTI_LOCK_WRITE);
    int rc = apply_join_state(&g_tc.c, w, &ep, &hconfig);
    unlock(w);

    TEST_ASSERT_EQUAL_INT(0, rc);
    TEST_ASSERT_EQUAL_FLOAT_MESSAGE(
        12.0f, get_worker_res_fval(w, "slots"),
        "mxj=-1 must auto-detect slots from the worker's reported maxcpus, "
        "not persist -1 or silently leave slots at 0");

    teardown_test_cluster();
    printf("test_apply_join_state_mxj_negative_one_falls_back_to_maxcpus: "
           "PASS\n");
}

/* ── Task 1.2: Tests for init_worker ──────────────────────────────────  */

/*
 * After init_worker for a new hostname, key_find must find the entry and
 * return a non-negative index.  Req 2.1, 2.3
 */
static void test_init_worker_creates_new_entry(void)
{
    setup_test_cluster();

    hostattrs_t attrs;
    memset(&attrs, 0, sizeof(attrs));
    attrs.hosttype  = makeString("LINUX");
    attrs.hostmodel = makeString("x86");
    attrs.cpufactor = 1.0f;
    attrs.maxcpus   = 4;
    attrs.maxmem    = 16;
    attrs.maxswap   = 32;
    attrs.maxtmp    = 64;
    attrs.ndisks    = 1;

    xce_hostconfig_t hconfig;
    memset(&hconfig, 0, sizeof(hconfig));
    hconfig.hosttype  = (char*)"!";
    hconfig.hostmodel = (char*)"!";
    hconfig.mxj       = 4;

    endpoint_t ep;
    memset(&ep, 0, sizeof(ep));

    int idx = init_worker(&g_tc.c, &ep, "new-host-1", &attrs, &hconfig);
    TEST_ASSERT_GREATER_OR_EQUAL_MESSAGE(0, idx,
                                         "init_worker must return non-negative index");

    worker_t* w = NULL;
    lock(g_tc.c.workermgr, RTTI_LOCK_READ);
    int rc = key_find(g_tc.c.workermgr, (void*)"new-host-1", (void**)&w);
    unlock(g_tc.c.workermgr);
    TEST_ASSERT_GREATER_THAN_MESSAGE(-1, rc,
                                     "worker must be registered in workermgr after init");
    TEST_ASSERT_NOT_NULL_MESSAGE(w, "key_find must return a valid worker pointer");

    teardown_test_cluster();
    printf("test_init_worker_creates_new_entry: PASS\n");
}

/*
 * The index returned by init_worker must match w->idx.  Req 2.3
 */
static void test_init_worker_returns_correct_idx(void)
{
    setup_test_cluster();

    hostattrs_t attrs;
    memset(&attrs, 0, sizeof(attrs));
    attrs.hosttype  = makeString("LINUX");
    attrs.hostmodel = makeString("x86");
    attrs.cpufactor = 1.0f;
    attrs.maxcpus   = 2;
    attrs.maxmem    = 8;
    attrs.maxswap   = 16;
    attrs.maxtmp    = 32;
    attrs.ndisks    = 1;

    xce_hostconfig_t hconfig;
    memset(&hconfig, 0, sizeof(hconfig));
    hconfig.hosttype  = (char*)"!";
    hconfig.hostmodel = (char*)"!";
    hconfig.mxj       = 2;

    endpoint_t ep;
    memset(&ep, 0, sizeof(ep));

    int idx = init_worker(&g_tc.c, &ep, "new-host-2", &attrs, &hconfig);
    TEST_ASSERT_GREATER_OR_EQUAL(0, idx);

    worker_t* w = NULL;
    lock(g_tc.c.workermgr, RTTI_LOCK_READ);
    key_find(g_tc.c.workermgr, (void*)"new-host-2", (void**)&w);
    unlock(g_tc.c.workermgr);
    TEST_ASSERT_NOT_NULL(w);
    TEST_ASSERT_EQUAL_INT_MESSAGE(idx, w->idx,
                                  "returned idx must match the worker's idx field");

    teardown_test_cluster();
    printf("test_init_worker_returns_correct_idx: PASS\n");
}

/* ── Task 1.3: Tests for rejoin_worker ────────────────────────────────  */

/*
 * rejoin_worker must update hostattrs and ep on the existing entry.  Req 3.1, 3.2
 */
static void test_rejoin_worker_updates_attrs_and_ep(void)
{
    setup_test_cluster();

    hostattrs_t old_attrs;
    memset(&old_attrs, 0, sizeof(old_attrs));
    old_attrs.hosttype  = makeString("OLD_TYPE");
    old_attrs.hostmodel = makeString("OLD_MODEL");
    old_attrs.cpufactor = 1.0f;
    old_attrs.maxcpus   = 2;
    old_attrs.maxmem    = 8;
    old_attrs.maxswap   = 16;
    old_attrs.maxtmp    = 32;
    old_attrs.ndisks    = 1;

    worker_t* w = make_test_worker("rejoin-host-1", &old_attrs);
    TEST_ASSERT_NOT_NULL(w);

    hostattrs_t new_attrs;
    memset(&new_attrs, 0, sizeof(new_attrs));
    new_attrs.hosttype  = makeString("NEW_TYPE");
    new_attrs.hostmodel = makeString("NEW_MODEL");
    new_attrs.cpufactor = 2.0f;
    new_attrs.maxcpus   = 8;
    new_attrs.maxmem    = 64;
    new_attrs.maxswap   = 128;
    new_attrs.maxtmp    = 256;
    new_attrs.ndisks    = 2;

    xce_hostconfig_t hconfig;
    memset(&hconfig, 0, sizeof(hconfig));
    hconfig.hosttype  = (char*)"!";
    hconfig.hostmodel = (char*)"!";
    hconfig.mxj       = 4;

    endpoint_t ep;
    memset(&ep, 0, sizeof(ep));

    int idx = rejoin_worker(&g_tc.c, w, &ep, &new_attrs, &hconfig);
    TEST_ASSERT_GREATER_OR_EQUAL_MESSAGE(0, idx,
                                         "rejoin_worker must return non-negative index");

    TEST_ASSERT_EQUAL_STRING_MESSAGE("NEW_TYPE", w->host.attrs->hosttype,
                                     "hosttype must be updated to new attrs value");
    TEST_ASSERT_EQUAL_STRING_MESSAGE("NEW_MODEL", w->host.attrs->hostmodel,
                                     "hostmodel must be updated to new attrs value");
    TEST_ASSERT_EQUAL_FLOAT_MESSAGE(2.0f, w->host.attrs->cpufactor,
                                    "cpufactor must be updated to new value");
    TEST_ASSERT_EQUAL_PTR_MESSAGE(&ep, w->ep,
                                  "ep must be updated to the new endpoint");

    /* Release the test's own references to the new strings */
    dereference(new_attrs.hosttype);
    dereference(new_attrs.hostmodel);

    teardown_test_cluster();
    printf("test_rejoin_worker_updates_attrs_and_ep: PASS\n");
}

/*
 * Admin-set status bits must survive a rejoin_worker call.  Req 3.3
 */
static void test_rejoin_worker_preserves_admin_bits(void)
{
    setup_test_cluster();

    hostattrs_t attrs;
    memset(&attrs, 0, sizeof(attrs));
    attrs.hosttype  = makeString("LINUX");
    attrs.hostmodel = makeString("x86");
    attrs.cpufactor = 1.0f;
    attrs.maxcpus   = 4;
    attrs.maxmem    = 16;
    attrs.maxswap   = 32;
    attrs.maxtmp    = 64;
    attrs.ndisks    = 1;

    worker_t* w = make_test_worker("rejoin-host-2", &attrs);
    TEST_ASSERT_NOT_NULL(w);

    /* Set all admin bits before the rejoin */
    lock(w, RTTI_LOCK_WRITE);
    w->host.status |= (XCE_WORKER_LOCKEDU | XCE_WORKER_CLOSED
                       | XCE_WORKER_CLOSED_BY_ADMIN);
    unlock(w);

    hostattrs_t new_attrs;
    memset(&new_attrs, 0, sizeof(new_attrs));
    new_attrs.hosttype  = makeString("LINUX");
    new_attrs.hostmodel = makeString("x86");
    new_attrs.cpufactor = 2.0f;
    new_attrs.maxcpus   = 8;
    new_attrs.maxmem    = 32;
    new_attrs.maxswap   = 64;
    new_attrs.maxtmp    = 128;
    new_attrs.ndisks    = 2;

    xce_hostconfig_t hconfig;
    memset(&hconfig, 0, sizeof(hconfig));
    hconfig.hosttype  = (char*)"!";
    hconfig.hostmodel = (char*)"!";
    hconfig.mxj       = 4;

    endpoint_t ep;
    memset(&ep, 0, sizeof(ep));

    rejoin_worker(&g_tc.c, w, &ep, &new_attrs, &hconfig);

    TEST_ASSERT_TRUE_MESSAGE(w->host.status & XCE_WORKER_LOCKEDU,
                             "LOCKEDU must be preserved after rejoin");
    TEST_ASSERT_TRUE_MESSAGE(w->host.status & XCE_WORKER_CLOSED,
                             "CLOSED must be preserved after rejoin");
    TEST_ASSERT_TRUE_MESSAGE(w->host.status & XCE_WORKER_CLOSED_BY_ADMIN,
                             "CLOSED_BY_ADMIN must be preserved after rejoin");

    dereference(new_attrs.hosttype);
    dereference(new_attrs.hostmodel);

    teardown_test_cluster();
    printf("test_rejoin_worker_preserves_admin_bits: PASS\n");
}

/* ── Dispatch ───────────────────────────────────────────────────────── */

void test_worker_join(void)
{
    UnitySetTestFile(__FILE__);
    printf("\n>> worker join refactor tests\n\n");

    /* xstring_table_init must be called once before any makeString usage */
    xstring_table_init(139);

    RUN_TEST(test_update_resvals_logs_actual_rc_not_stale_errno);

    /* Task 1.1: apply_join_state (shared status/resource update) */
    RUN_TEST(test_apply_join_state_clears_unavail);
    RUN_TEST(test_apply_join_state_applies_hosttype_override);
    RUN_TEST(test_apply_join_state_updates_resource_values);
    RUN_TEST(test_apply_join_state_mxj_negative_one_falls_back_to_maxcpus);

    /* Task 1.2: init_worker (first-time join) */
    RUN_TEST(test_init_worker_creates_new_entry);
    RUN_TEST(test_init_worker_returns_correct_idx);

    /* Task 1.3: rejoin_worker (reconnect with admin-bit preservation) */
    RUN_TEST(test_rejoin_worker_updates_attrs_and_ep);
    RUN_TEST(test_rejoin_worker_preserves_admin_bits);
}