feat: JOB_STAT_TRANSIENT, job_should_retain, worker_should_retain, dead_since (P0-5a, P0-8, P0-13)
Add terminal-state constants and GC retention predicates for the checkpoint xzarray_thaw() GC pass: - types.h: add JOB_STAT_TRANSIENT (0x400), XCE_WORKER_HIDDEN (0x00200000), QUEUE_STAT_HIDDEN (0x0100) - persist.c: implement job_is_terminal(), job_should_retain(), worker_should_retain() — placed here so the unit test binary (which links only persist.c) can access them without RTTI deps - persist.h: declare the three GC predicates; include xce/types.h - worker.c: set dead_since = time(NULL) in mark_worker_unavail() when dead_since is not already set - cluster.c: add TODO P0-8b marker on zarray_del endpoint cleanup - meson.build: register test_gc_retention.c; raise test timeout 60→120s - testsuite/test_gc_retention.c: 19 tests covering job_is_terminal, job_should_retain (TRANSIENT TTL, terminal TTL, non-terminal always retained), and worker_should_retain (HIDDEN/UNAVAIL/healthy/zero dead_since); all pass (63 total, 0 failures)
This commit is contained in:
parent
6f3b98c6b5
commit
3030012f51
9 changed files with 330 additions and 2 deletions
|
|
@ -12,6 +12,7 @@ extern void test_persist_checkpoint(void);
|
|||
extern void test_persist_recovery(void);
|
||||
extern void test_hooks(void);
|
||||
extern void test_persist_hooks(void);
|
||||
extern void test_gc_retention(void);
|
||||
|
||||
void setUp(void) { return; }
|
||||
|
||||
|
|
@ -93,6 +94,7 @@ int main(int argc, char* argv[])
|
|||
TEST(test_persist_recovery);
|
||||
TEST(test_hooks);
|
||||
TEST(test_persist_hooks);
|
||||
TEST(test_gc_retention);
|
||||
|
||||
return UNITY_END();
|
||||
}
|
||||
|
|
|
|||
244
src/controller/testsuite/test_gc_retention.c
Normal file
244
src/controller/testsuite/test_gc_retention.c
Normal file
|
|
@ -0,0 +1,244 @@
|
|||
#define _GNU_SOURCE
|
||||
#include <stdbool.h>
|
||||
#include <stdio.h>
|
||||
#include <string.h>
|
||||
#include <time.h>
|
||||
|
||||
#include <unity/unity.h>
|
||||
|
||||
#include "config.h"
|
||||
#include "persist.h"
|
||||
#include "xce/types.h"
|
||||
|
||||
/* ------------------------------------------------------------------ */
|
||||
/* Helpers to build minimal job_t / xce_worker_t fixtures */
|
||||
/* ------------------------------------------------------------------ */
|
||||
|
||||
static job_t make_job(unsigned int state, time_t finish_time,
|
||||
uint64_t transient_since)
|
||||
{
|
||||
job_t j;
|
||||
memset(&j, 0, sizeof(j));
|
||||
j.job.state = state;
|
||||
j.job.finish_time = finish_time;
|
||||
j.job.transient_since = (uint64_t)transient_since;
|
||||
return j;
|
||||
}
|
||||
|
||||
static xce_worker_t make_worker(uint32_t status, time_t dead_since)
|
||||
{
|
||||
xce_worker_t w;
|
||||
memset(&w, 0, sizeof(w));
|
||||
w.status = status;
|
||||
w.dead_since = dead_since;
|
||||
return w;
|
||||
}
|
||||
|
||||
/* ------------------------------------------------------------------ */
|
||||
/* Forward declarations for functions under test */
|
||||
/* ------------------------------------------------------------------ */
|
||||
|
||||
extern bool job_is_terminal(const job_t* j);
|
||||
extern bool job_should_retain(const job_t* j, const persist_config_t* cfg);
|
||||
extern bool worker_should_retain(const xce_worker_t* w,
|
||||
const persist_config_t* cfg);
|
||||
|
||||
/* ------------------------------------------------------------------ */
|
||||
/* job_is_terminal tests */
|
||||
/* ------------------------------------------------------------------ */
|
||||
|
||||
static void test_running_job_not_terminal(void)
|
||||
{
|
||||
job_t j = make_job(JOB_STAT_RUN, 0, 0);
|
||||
TEST_ASSERT_FALSE(job_is_terminal(&j));
|
||||
}
|
||||
|
||||
static void test_pending_job_not_terminal(void)
|
||||
{
|
||||
job_t j = make_job(JOB_STAT_PEND, 0, 0);
|
||||
TEST_ASSERT_FALSE(job_is_terminal(&j));
|
||||
}
|
||||
|
||||
static void test_exit_job_is_terminal(void)
|
||||
{
|
||||
job_t j = make_job(JOB_STAT_EXIT, time(NULL), 0);
|
||||
TEST_ASSERT_TRUE(job_is_terminal(&j));
|
||||
}
|
||||
|
||||
static void test_done_job_is_terminal(void)
|
||||
{
|
||||
job_t j = make_job(JOB_STAT_DONE, time(NULL), 0);
|
||||
TEST_ASSERT_TRUE(job_is_terminal(&j));
|
||||
}
|
||||
|
||||
static void test_perr_job_is_terminal(void)
|
||||
{
|
||||
job_t j = make_job(JOB_STAT_PERR, time(NULL), 0);
|
||||
TEST_ASSERT_TRUE(job_is_terminal(&j));
|
||||
}
|
||||
|
||||
static void test_transient_job_not_terminal(void)
|
||||
{
|
||||
/* JOB_STAT_TRANSIENT is NOT terminal — it's a special non-terminal state */
|
||||
job_t j = make_job(JOB_STAT_TRANSIENT, 0, (uint64_t)time(NULL));
|
||||
TEST_ASSERT_FALSE(job_is_terminal(&j));
|
||||
}
|
||||
|
||||
/* ------------------------------------------------------------------ */
|
||||
/* job_should_retain: TRANSIENT timeout tests */
|
||||
/* ------------------------------------------------------------------ */
|
||||
|
||||
static void test_transient_job_retained_within_timeout(void)
|
||||
{
|
||||
/* transient_since = now - 100s, timeout = 300s → retain */
|
||||
time_t now = time(NULL);
|
||||
persist_config_t cfg = persist_config_defaults();
|
||||
cfg.transient_job_timeout_s = 300;
|
||||
job_t j = make_job(JOB_STAT_TRANSIENT, 0, now - 100);
|
||||
TEST_ASSERT_TRUE(job_should_retain(&j, &cfg));
|
||||
}
|
||||
|
||||
static void test_transient_job_expired(void)
|
||||
{
|
||||
/* transient_since = now - 400s, timeout = 300s → do NOT retain */
|
||||
time_t now = time(NULL);
|
||||
persist_config_t cfg = persist_config_defaults();
|
||||
cfg.transient_job_timeout_s = 300;
|
||||
job_t j = make_job(JOB_STAT_TRANSIENT, 0, now - 400);
|
||||
TEST_ASSERT_FALSE(job_should_retain(&j, &cfg));
|
||||
}
|
||||
|
||||
static void test_transient_job_zero_since_always_retained(void)
|
||||
{
|
||||
/* transient_since == 0 means unset → never expire (operator must decide) */
|
||||
persist_config_t cfg = persist_config_defaults();
|
||||
cfg.transient_job_timeout_s = 300;
|
||||
job_t j = make_job(JOB_STAT_TRANSIENT, 0, 0);
|
||||
TEST_ASSERT_TRUE(job_should_retain(&j, &cfg));
|
||||
}
|
||||
|
||||
/* ------------------------------------------------------------------ */
|
||||
/* job_should_retain: terminal job TTL tests */
|
||||
/* ------------------------------------------------------------------ */
|
||||
|
||||
static void test_finished_job_retained_within_ttl(void)
|
||||
{
|
||||
/* finish_time = now - 1800s, retain = 3600s → retain */
|
||||
time_t now = time(NULL);
|
||||
persist_config_t cfg = persist_config_defaults();
|
||||
cfg.finished_retain_s = 3600;
|
||||
job_t j = make_job(JOB_STAT_EXIT, now - 1800, 0);
|
||||
TEST_ASSERT_TRUE(job_should_retain(&j, &cfg));
|
||||
}
|
||||
|
||||
static void test_finished_job_expired(void)
|
||||
{
|
||||
/* finish_time = now - 7200s, retain = 3600s → do NOT retain */
|
||||
time_t now = time(NULL);
|
||||
persist_config_t cfg = persist_config_defaults();
|
||||
cfg.finished_retain_s = 3600;
|
||||
job_t j = make_job(JOB_STAT_EXIT, now - 7200, 0);
|
||||
TEST_ASSERT_FALSE(job_should_retain(&j, &cfg));
|
||||
}
|
||||
|
||||
static void test_done_job_retained_within_ttl(void)
|
||||
{
|
||||
time_t now = time(NULL);
|
||||
persist_config_t cfg = persist_config_defaults();
|
||||
cfg.finished_retain_s = 3600;
|
||||
job_t j = make_job(JOB_STAT_DONE, now - 1800, 0);
|
||||
TEST_ASSERT_TRUE(job_should_retain(&j, &cfg));
|
||||
}
|
||||
|
||||
static void test_running_job_always_retained(void)
|
||||
{
|
||||
persist_config_t cfg = persist_config_defaults();
|
||||
job_t j = make_job(JOB_STAT_RUN, 0, 0);
|
||||
TEST_ASSERT_TRUE(job_should_retain(&j, &cfg));
|
||||
}
|
||||
|
||||
static void test_pending_job_always_retained(void)
|
||||
{
|
||||
persist_config_t cfg = persist_config_defaults();
|
||||
job_t j = make_job(JOB_STAT_PEND, 0, 0);
|
||||
TEST_ASSERT_TRUE(job_should_retain(&j, &cfg));
|
||||
}
|
||||
|
||||
/* ------------------------------------------------------------------ */
|
||||
/* worker_should_retain tests */
|
||||
/* ------------------------------------------------------------------ */
|
||||
|
||||
static void test_worker_unavail_retained_within_ttl(void)
|
||||
{
|
||||
/* dead_since = now - 43200s (12 h), retain = 86400s (24 h) → retain */
|
||||
time_t now = time(NULL);
|
||||
persist_config_t cfg = persist_config_defaults();
|
||||
cfg.worker_retain_s = 86400;
|
||||
xce_worker_t w = make_worker(XCE_WORKER_UNAVAIL, now - 43200);
|
||||
TEST_ASSERT_TRUE(worker_should_retain(&w, &cfg));
|
||||
}
|
||||
|
||||
static void test_worker_unavail_expired(void)
|
||||
{
|
||||
/* dead_since = now - 100000s, retain = 86400s → do NOT retain */
|
||||
time_t now = time(NULL);
|
||||
persist_config_t cfg = persist_config_defaults();
|
||||
cfg.worker_retain_s = 86400;
|
||||
xce_worker_t w = make_worker(XCE_WORKER_UNAVAIL, now - 100000);
|
||||
TEST_ASSERT_FALSE(worker_should_retain(&w, &cfg));
|
||||
}
|
||||
|
||||
static void test_worker_hidden_never_retained(void)
|
||||
{
|
||||
/* XCE_WORKER_HIDDEN → cleanup immediately regardless of dead_since */
|
||||
time_t now = time(NULL);
|
||||
persist_config_t cfg = persist_config_defaults();
|
||||
xce_worker_t w = make_worker(XCE_WORKER_HIDDEN, now - 10);
|
||||
TEST_ASSERT_FALSE(worker_should_retain(&w, &cfg));
|
||||
}
|
||||
|
||||
static void test_worker_healthy_always_retained(void)
|
||||
{
|
||||
persist_config_t cfg = persist_config_defaults();
|
||||
xce_worker_t w = make_worker(0, 0);
|
||||
TEST_ASSERT_TRUE(worker_should_retain(&w, &cfg));
|
||||
}
|
||||
|
||||
static void test_worker_unavail_zero_dead_since_always_retained(void)
|
||||
{
|
||||
/* dead_since == 0 → not yet set, do not GC */
|
||||
persist_config_t cfg = persist_config_defaults();
|
||||
cfg.worker_retain_s = 86400;
|
||||
xce_worker_t w = make_worker(XCE_WORKER_UNAVAIL, 0);
|
||||
TEST_ASSERT_TRUE(worker_should_retain(&w, &cfg));
|
||||
}
|
||||
|
||||
/* ------------------------------------------------------------------ */
|
||||
/* Test suite entry point */
|
||||
/* ------------------------------------------------------------------ */
|
||||
|
||||
void test_gc_retention(void)
|
||||
{
|
||||
RUN_TEST(test_running_job_not_terminal);
|
||||
RUN_TEST(test_pending_job_not_terminal);
|
||||
RUN_TEST(test_exit_job_is_terminal);
|
||||
RUN_TEST(test_done_job_is_terminal);
|
||||
RUN_TEST(test_perr_job_is_terminal);
|
||||
RUN_TEST(test_transient_job_not_terminal);
|
||||
|
||||
RUN_TEST(test_transient_job_retained_within_timeout);
|
||||
RUN_TEST(test_transient_job_expired);
|
||||
RUN_TEST(test_transient_job_zero_since_always_retained);
|
||||
|
||||
RUN_TEST(test_finished_job_retained_within_ttl);
|
||||
RUN_TEST(test_finished_job_expired);
|
||||
RUN_TEST(test_done_job_retained_within_ttl);
|
||||
RUN_TEST(test_running_job_always_retained);
|
||||
RUN_TEST(test_pending_job_always_retained);
|
||||
|
||||
RUN_TEST(test_worker_unavail_retained_within_ttl);
|
||||
RUN_TEST(test_worker_unavail_expired);
|
||||
RUN_TEST(test_worker_hidden_never_retained);
|
||||
RUN_TEST(test_worker_healthy_always_retained);
|
||||
RUN_TEST(test_worker_unavail_zero_dead_since_always_retained);
|
||||
}
|
||||
Loading…
Add table
Add a link
Reference in a new issue