fix(worker): real maxswap/maxtmp detection; feat(scheduler): annotate multi-slot assigned jobs
bug-1554: init_host_attrs() replaces hardcoded placeholder stubs with real
detection -- maxswap from sysinfo()'s totalswap (reusing the same syscall
snapshot already taken for maxmem), maxtmp from a fresh statvfs("/tmp")
total-capacity measurement. No synthetic fallback on detection failure, so
a system with swap genuinely disabled correctly reports 0.
bug-1776: the scheduler simulator's "info worker" assigned-jobs listing now
annotates a job consuming more than one "slots" unit (e.g. "jobid*2"),
using per-host allocation data that was already being walked but discarded.
Falls back to the pre-existing per-job listing when jobs on the same worker
have differing slot counts, since a single suffix on a compressed range
would be ambiguous.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Hb6Edtj4KbS8CrgEWYdA2E
This commit is contained in:
parent
a6b735b6f2
commit
a35f597e06
9 changed files with 660 additions and 28 deletions
|
|
@ -1056,6 +1056,245 @@ test_info_worker_assigned_job_listing_compresses_contiguous_array_indices(
|
|||
"array_indices: PASS\n");
|
||||
}
|
||||
|
||||
/*
|
||||
* Improvement request (simulator REPL user feedback): a job consuming more
|
||||
* than one "slots" unit on a worker should be visibly distinguishable from
|
||||
* a single-slot job in the assigned-jobs listing (e.g. "jobid*2"), instead
|
||||
* of every assigned job looking identical regardless of how much of the
|
||||
* worker it actually occupies.
|
||||
*/
|
||||
static void
|
||||
test_info_worker_assigned_job_listing_appends_slot_count_when_uniform(void)
|
||||
{
|
||||
xce_sched_sim_driver_t sim;
|
||||
xce_core_event_t events[2];
|
||||
xce_queue_t wire_queue;
|
||||
xce_worker_t wire_worker;
|
||||
xce_worker_t* worker_ptrs[1];
|
||||
xce_job_t* job;
|
||||
xce_job_t* all_jobs[1];
|
||||
uint64_t jobid;
|
||||
xce_sched_job_t* sjob;
|
||||
xce_alloc_t slots_alloc;
|
||||
xce_alloc_t* slots_alloc_ptrs[1];
|
||||
xce_host_alloc_t ha_host;
|
||||
xce_host_alloc_t* ha_host_ptrs[1];
|
||||
xce_job_alloc_t alloc;
|
||||
char* captured;
|
||||
char expected_range[256];
|
||||
char expected_line[300];
|
||||
int rc;
|
||||
|
||||
memset(&sim, 0, sizeof(sim));
|
||||
rc = xce_sched_core_init(&(sim.core));
|
||||
TEST_ASSERT_EQUAL_INT(0, rc);
|
||||
|
||||
memset(&wire_queue, 0, sizeof(wire_queue));
|
||||
wire_queue.queue = "normal";
|
||||
wire_queue.priority = 5;
|
||||
wire_queue.status = QUEUE_STAT_OPEN;
|
||||
|
||||
jobid = mk_plain_jobid_bpi(4, 2, 1700000010, 1);
|
||||
job = make_job(jobid, JOB_STAT_RUN, "normal", "alice", "proj1");
|
||||
all_jobs[0] = job;
|
||||
|
||||
memset(events, 0, sizeof(events));
|
||||
events[0].type = XCE_CORE_EVENT_QUEUE_UPDATE;
|
||||
events[0].u.queue_update.queue = &wire_queue;
|
||||
events[1].type = XCE_CORE_EVENT_JOBS_PUSH;
|
||||
events[1].u.jobs_push.njobs = 1;
|
||||
events[1].u.jobs_push.jobs = all_jobs;
|
||||
|
||||
rc = xce_core_push_events(&(sim.core), events, 2);
|
||||
TEST_ASSERT_EQUAL_INT(0, rc);
|
||||
|
||||
memset(&wire_worker, 0, sizeof(wire_worker));
|
||||
wire_worker.name = "hostA";
|
||||
wire_worker.status = 0;
|
||||
worker_ptrs[0] = &wire_worker;
|
||||
|
||||
memset(events, 0, sizeof(events));
|
||||
events[0].type = XCE_CORE_EVENT_WORKERS_PUSH;
|
||||
events[0].u.workers_push.nworkers = 1;
|
||||
events[0].u.workers_push.workers = worker_ptrs;
|
||||
|
||||
rc = xce_core_push_events(&(sim.core), events, 1);
|
||||
TEST_ASSERT_EQUAL_INT(0, rc);
|
||||
|
||||
memset(&slots_alloc, 0, sizeof(slots_alloc));
|
||||
slots_alloc.resname = "slots";
|
||||
slots_alloc.value = 2.0;
|
||||
slots_alloc_ptrs[0] = &slots_alloc;
|
||||
|
||||
memset(&ha_host, 0, sizeof(ha_host));
|
||||
ha_host.host = "hostA";
|
||||
ha_host.nallocs = 1;
|
||||
ha_host.allocs = slots_alloc_ptrs;
|
||||
ha_host_ptrs[0] = &ha_host;
|
||||
|
||||
memset(&alloc, 0, sizeof(alloc));
|
||||
alloc.nhosts = 1;
|
||||
alloc.allocs = ha_host_ptrs;
|
||||
|
||||
sjob = sched_get_job(&(sim.core), jobid);
|
||||
TEST_ASSERT_NOT_NULL(sjob);
|
||||
sjob->allocs = &alloc;
|
||||
|
||||
rc = repl_fmt_jobid_range(expected_range, sizeof(expected_range), &jobid,
|
||||
1);
|
||||
TEST_ASSERT_TRUE(rc >= 0);
|
||||
snprintf(expected_line, sizeof(expected_line), "assigned jobs: %s*2\n",
|
||||
expected_range);
|
||||
|
||||
captured = capture_dispatch(&sim, "info worker hostA");
|
||||
|
||||
TEST_ASSERT_NOT_NULL_MESSAGE(
|
||||
strstr(captured, expected_line),
|
||||
"a job consuming 2 'slots' on this worker must be annotated "
|
||||
"'*2' in the assigned-jobs listing");
|
||||
|
||||
free(captured);
|
||||
|
||||
printf("test_info_worker_assigned_job_listing_appends_slot_count_when_"
|
||||
"uniform: PASS\n");
|
||||
}
|
||||
|
||||
/*
|
||||
* Two jobs on the SAME worker consuming DIFFERENT slot counts cannot be
|
||||
* annotated with a single trailing "*N" on a compressed range (which count
|
||||
* would it even mean?) -- must fall back to the pre-existing uncompressed
|
||||
* per-job listing, each line annotated with its own count.
|
||||
*/
|
||||
static void
|
||||
test_info_worker_assigned_job_listing_falls_back_per_job_when_slots_differ(
|
||||
void)
|
||||
{
|
||||
xce_sched_sim_driver_t sim;
|
||||
xce_core_event_t events[2];
|
||||
xce_queue_t wire_queue;
|
||||
xce_worker_t wire_worker;
|
||||
xce_worker_t* worker_ptrs[1];
|
||||
xce_job_t* job_a;
|
||||
xce_job_t* job_b;
|
||||
xce_job_t* all_jobs[2];
|
||||
uint64_t jobid_a;
|
||||
uint64_t jobid_b;
|
||||
xce_sched_job_t* sjob;
|
||||
xce_alloc_t slots_alloc_a;
|
||||
xce_alloc_t* slots_alloc_a_ptrs[1];
|
||||
xce_alloc_t slots_alloc_b;
|
||||
xce_alloc_t* slots_alloc_b_ptrs[1];
|
||||
xce_host_alloc_t ha_a;
|
||||
xce_host_alloc_t* ha_a_ptrs[1];
|
||||
xce_job_alloc_t alloc_a;
|
||||
xce_host_alloc_t ha_b;
|
||||
xce_host_alloc_t* ha_b_ptrs[1];
|
||||
xce_job_alloc_t alloc_b;
|
||||
char* captured;
|
||||
char jobidStr_a[64];
|
||||
char jobidStr_b[64];
|
||||
char expected_a[96];
|
||||
char expected_b[96];
|
||||
int rc;
|
||||
|
||||
memset(&sim, 0, sizeof(sim));
|
||||
rc = xce_sched_core_init(&(sim.core));
|
||||
TEST_ASSERT_EQUAL_INT(0, rc);
|
||||
|
||||
memset(&wire_queue, 0, sizeof(wire_queue));
|
||||
wire_queue.queue = "normal";
|
||||
wire_queue.priority = 5;
|
||||
wire_queue.status = QUEUE_STAT_OPEN;
|
||||
|
||||
jobid_a = mk_plain_jobid_bpi(5, 2, 1700000020, 1);
|
||||
jobid_b = mk_plain_jobid_bpi(5, 2, 1700000021, 2);
|
||||
job_a = make_job(jobid_a, JOB_STAT_RUN, "normal", "alice", "proj1");
|
||||
job_b = make_job(jobid_b, JOB_STAT_RUN, "normal", "alice", "proj1");
|
||||
all_jobs[0] = job_a;
|
||||
all_jobs[1] = job_b;
|
||||
|
||||
memset(events, 0, sizeof(events));
|
||||
events[0].type = XCE_CORE_EVENT_QUEUE_UPDATE;
|
||||
events[0].u.queue_update.queue = &wire_queue;
|
||||
events[1].type = XCE_CORE_EVENT_JOBS_PUSH;
|
||||
events[1].u.jobs_push.njobs = 2;
|
||||
events[1].u.jobs_push.jobs = all_jobs;
|
||||
|
||||
rc = xce_core_push_events(&(sim.core), events, 2);
|
||||
TEST_ASSERT_EQUAL_INT(0, rc);
|
||||
|
||||
memset(&wire_worker, 0, sizeof(wire_worker));
|
||||
wire_worker.name = "hostA";
|
||||
wire_worker.status = 0;
|
||||
worker_ptrs[0] = &wire_worker;
|
||||
|
||||
memset(events, 0, sizeof(events));
|
||||
events[0].type = XCE_CORE_EVENT_WORKERS_PUSH;
|
||||
events[0].u.workers_push.nworkers = 1;
|
||||
events[0].u.workers_push.workers = worker_ptrs;
|
||||
|
||||
rc = xce_core_push_events(&(sim.core), events, 1);
|
||||
TEST_ASSERT_EQUAL_INT(0, rc);
|
||||
|
||||
memset(&slots_alloc_a, 0, sizeof(slots_alloc_a));
|
||||
slots_alloc_a.resname = "slots";
|
||||
slots_alloc_a.value = 2.0;
|
||||
slots_alloc_a_ptrs[0] = &slots_alloc_a;
|
||||
memset(&ha_a, 0, sizeof(ha_a));
|
||||
ha_a.host = "hostA";
|
||||
ha_a.nallocs = 1;
|
||||
ha_a.allocs = slots_alloc_a_ptrs;
|
||||
ha_a_ptrs[0] = &ha_a;
|
||||
memset(&alloc_a, 0, sizeof(alloc_a));
|
||||
alloc_a.nhosts = 1;
|
||||
alloc_a.allocs = ha_a_ptrs;
|
||||
|
||||
memset(&slots_alloc_b, 0, sizeof(slots_alloc_b));
|
||||
slots_alloc_b.resname = "slots";
|
||||
slots_alloc_b.value = 4.0;
|
||||
slots_alloc_b_ptrs[0] = &slots_alloc_b;
|
||||
memset(&ha_b, 0, sizeof(ha_b));
|
||||
ha_b.host = "hostA";
|
||||
ha_b.nallocs = 1;
|
||||
ha_b.allocs = slots_alloc_b_ptrs;
|
||||
ha_b_ptrs[0] = &ha_b;
|
||||
memset(&alloc_b, 0, sizeof(alloc_b));
|
||||
alloc_b.nhosts = 1;
|
||||
alloc_b.allocs = ha_b_ptrs;
|
||||
|
||||
sjob = sched_get_job(&(sim.core), jobid_a);
|
||||
TEST_ASSERT_NOT_NULL(sjob);
|
||||
sjob->allocs = &alloc_a;
|
||||
sjob = sched_get_job(&(sim.core), jobid_b);
|
||||
TEST_ASSERT_NOT_NULL(sjob);
|
||||
sjob->allocs = &alloc_b;
|
||||
|
||||
fmt_jobid(jobidStr_a, sizeof(jobidStr_a), jobid_a);
|
||||
fmt_jobid(jobidStr_b, sizeof(jobidStr_b), jobid_b);
|
||||
snprintf(expected_a, sizeof(expected_a), "assigned job: jobid=%s*2\n",
|
||||
jobidStr_a);
|
||||
snprintf(expected_b, sizeof(expected_b), "assigned job: jobid=%s*4\n",
|
||||
jobidStr_b);
|
||||
|
||||
captured = capture_dispatch(&sim, "info worker hostA");
|
||||
|
||||
TEST_ASSERT_NOT_NULL_MESSAGE(
|
||||
strstr(captured, expected_a),
|
||||
"differing per-job slot counts on the same worker must fall back "
|
||||
"to the uncompressed per-job listing, annotated with this job's "
|
||||
"own slot count");
|
||||
TEST_ASSERT_NOT_NULL_MESSAGE(
|
||||
strstr(captured, expected_b),
|
||||
"differing per-job slot counts on the same worker must fall back "
|
||||
"to the uncompressed per-job listing, annotated with this job's "
|
||||
"own slot count");
|
||||
|
||||
free(captured);
|
||||
|
||||
printf("test_info_worker_assigned_job_listing_falls_back_per_job_when_"
|
||||
"slots_differ: PASS\n");
|
||||
}
|
||||
|
||||
/*
|
||||
* AC 6.6, `info resource all`: the same three-section grouping as `info
|
||||
* worker`'s own version above, applied to the cluster-wide GLOBAL/SHARED
|
||||
|
|
@ -1366,6 +1605,8 @@ int main(void)
|
|||
RUN_TEST(test_info_worker_units_mode_both_renders_raw_and_human_without_mutating_stored_value);
|
||||
RUN_TEST(test_info_worker_units_mode_converts_swp_megabytes_before_human_scaling);
|
||||
RUN_TEST(test_info_worker_assigned_job_listing_compresses_contiguous_array_indices);
|
||||
RUN_TEST(test_info_worker_assigned_job_listing_appends_slot_count_when_uniform);
|
||||
RUN_TEST(test_info_worker_assigned_job_listing_falls_back_per_job_when_slots_differ);
|
||||
RUN_TEST(test_info_resource_groups_resources_into_labeled_table_sections_in_order);
|
||||
RUN_TEST(test_info_resource_load_metric_whole_number_omits_trailing_zeros);
|
||||
RUN_TEST(test_info_resource_units_mode_both_renders_raw_and_human_for_bytes_resource);
|
||||
|
|
|
|||
Loading…
Add table
Add a link
Reference in a new issue