lei.chen/myxce
1
0
Fork 1
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

feat(resreq-maxtasks-wiring): wire fresh job-ingest to real task count

Browse source 
core_build_job() now calls apply_job_task_count() right after its
resreq parse succeeds and before core_build_schedobj() reads the
resreq's task count for resreq_fp — so a freshly-submitted multi-task
job's dispatched primary resreq reflects its real requested min/max
task count instead of the previous always-one-task hardcode.

Tracked follow-up (not blocking, see tasks.md Implementation Notes):
test_repl_breakpoints_inspection.c:735 hardcodes an incidental
"Tasks: min = 0, max = 0" byproduct of a fixture's parse_select()
failure; now correctly reports (1,1) per Requirement 1.2. Fix folded
into task 5.1's full-suite regression pass.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XYdtrEWWZfKRRGED7raEYg
This commit is contained in:
Rongsong Shen 2026-09-06 04:01:23 +00:00
parent 799a71ecb1
commit 0aadb49493
No known key found for this signature in database
GPG key ID: 5BCDAAE0123E9FE2
3 changed files with 107 additions and 1 deletions
Download patch file Download diff file Expand all files Collapse all files

5
.kiro/specs/resreq-maxtasks-wiring/tasks.md
View file

@ -10,7 +10,7 @@
- _Boundary: TaskCountWiring_
- [ ] 2. Core: wire the derived task count into live scheduler ingest paths and close the adjacent worker launch-type gap
- [ ] 2.1 (P) Wire fresh job-ingest to apply the job's real requested task count
- [x] 2.1 (P) Wire fresh job-ingest to apply the job's real requested task count
- Call the new helper immediately after fresh job ingest's existing resource-request parse, using the job's real submitted minimum/maximum task count already in scope there
- Observable: a job's dispatched resource request (the one the allocator actually reads) reflects the real requested task count immediately after fresh ingest, not the previous always-one-task default
- _Requirements: 1.1_
@ -87,3 +87,6 @@
- Re-run the existing sequential and chaotic HA soak unmodified and confirm no new job loss, divergence, or liveness stall attributable to this feature
- Observable: a clean soak pass recorded as this feature's HA non-regression evidence
- _Requirements: 5.2_
## Implementation Notes
- Task 2.1 surfaced an out-of-boundary side effect: `src/scheduler/testsuite/test_repl_breakpoints_inspection.c:735`'s `test_info_job_reports_state_queue_priority` hardcodes an incidental `"Tasks: min = 0, max = 0"` byproduct of its fixture's `parse_select()` failing on an unregistered `"type"` resource (which previously left the primary's task-count fields at memset-zero, before `apply_job_task_count()` existed). Now correctly overwritten to `(1,1)` per Requirement 1.2. Before task 5.1 is considered complete, change that assertion's expected substring to `"Tasks: min = 1, max = 1"`.

17
src/scheduler/src/core_events.c
View file

@ -256,6 +256,23 @@ static xce_sched_job_t* core_build_job(xce_sched_core_t* core, xce_job_t* j)
return NULL;
}
/* resreq-maxtasks-wiring task 2.1 (bug-1770, Req 1.1/1.2/1.5): the
* just-parsed resreq's standalone primary carries whatever
* parse_resreq_primary() left it at -- normally its own minTasks==
* maxTasks==1 hardcode, but NOT unconditionally (a standalone primary
* whose select clause fails to resolve against this core's resmgr
* returns early and leaves minTasks/maxTasks at 0, e.g. some fixtures'
* "type == any" against an empty resmgr) -- never the job's real
* requested count either way. Overwrite it here unconditionally, using
* the job's real submitted task count (j->submit->mintasks/maxtasks) --
* the SAME source job->mintasks/maxtasks below is populated from, just
* applied before core_build_schedobj() so the fingerprint it computes
* off job->resreq (core_build_primary_schedobj()'s resreq_fp) reflects
* the real value too, not whatever the just-parsed default happened to
* be. No-ops for a COMPOUND/ALTER resreq (apply_job_task_count()'s own
* tag gate) -- out of this task's boundary per design.md. */
apply_job_task_count(job->resreq, j->submit->mintasks, j->submit->maxtasks);
job->schedobj = (xce_sched_obj_t*)calloc(1, sizeof(*(job->schedobj)));
if (job->schedobj == NULL) {
FREEUP(job->resreq);

86
src/scheduler/testsuite/test_core_events.c
View file

@ -1545,6 +1545,90 @@ test_core_build_primary_schedobj_populates_resreq_fp_from_parsed_resreq(void)
FREEUP(errmsg);
}
/*
* resreq-maxtasks-wiring task 2.1 (bug-1770, Req 1.1): a job's real
* requested task count (job_submit_data_t.mintasks/maxtasks, already
* correctly threaded onto xce_job_t.submit by the time a JOBS_PUSH event
* reaches core_build_job()) must reach the standalone primary resreq the
* allocator actually dispatches on (job->resreq->u.primary->minTasks/
* maxTasks) -- not parse_resreq_primary()'s always-1 hardcoded default for
* every freshly-parsed resreq. Before this fix, core_build_job() never
* applied the new apply_job_task_count() helper (src/lib/resreq.c, task
* 1), so a job submitted with e.g. -n 4,6 silently ingested with
* minTasks==maxTasks==1, permanently unreachable by scheduler-allocation's
* parjob_allocator.
*/
static void
test_core_build_job_wires_real_task_count_into_resreq_primary(void)
{
xce_sched_core_t core;
xce_core_event_t event;
xce_job_t* job1;
xce_sched_job_t* tracked;
int rc;
fixture_init(&core, &g_full_plugin);
job1 = make_job(4160, JOB_STAT_PEND, "normal", "alice", "proj1");
job1->submit->mintasks = 4;
job1->submit->maxtasks = 6;
memset(&event, 0, sizeof(event));
event.type = XCE_CORE_EVENT_JOBS_PUSH;
event.u.jobs_push.njobs = 1;
event.u.jobs_push.jobs = &job1;
rc = xce_core_push_events(&core, &event, 1);
TEST_ASSERT_EQUAL_INT(0, rc);
tracked = sched_get_job(&core, 4160);
TEST_ASSERT_NOT_NULL(tracked);
TEST_ASSERT_NOT_NULL(tracked->resreq);
TEST_ASSERT_EQUAL_INT(XCE_RESREQ_PRIMARY, tracked->resreq->tag);
TEST_ASSERT_NOT_NULL(tracked->resreq->u.primary);
/* The fix under test: the standalone primary's minTasks/maxTasks must
* reflect the job's real submitted task count, not parse_resreq_
* primary()'s always-1 default. */
TEST_ASSERT_EQUAL_INT(4, tracked->resreq->u.primary->minTasks);
TEST_ASSERT_EQUAL_INT(6, tracked->resreq->u.primary->maxTasks);
}
/* resreq-maxtasks-wiring task 2.1 (Req 1.2): a job submitted without an
* explicit task count (make_job()'s default mintasks==maxtasks==1) must
* continue to ingest with minTasks==maxTasks==1 -- zero behavioral change
* for the common single-task case. */
static void
test_core_build_job_default_task_count_still_one(void)
{
xce_sched_core_t core;
xce_core_event_t event;
xce_job_t* job1;
xce_sched_job_t* tracked;
int rc;
fixture_init(&core, &g_full_plugin);
job1 = make_job(4161, JOB_STAT_PEND, "normal", "alice", "proj1");
memset(&event, 0, sizeof(event));
event.type = XCE_CORE_EVENT_JOBS_PUSH;
event.u.jobs_push.njobs = 1;
event.u.jobs_push.jobs = &job1;
rc = xce_core_push_events(&core, &event, 1);
TEST_ASSERT_EQUAL_INT(0, rc);
tracked = sched_get_job(&core, 4161);
TEST_ASSERT_NOT_NULL(tracked);
TEST_ASSERT_NOT_NULL(tracked->resreq);
TEST_ASSERT_EQUAL_INT(XCE_RESREQ_PRIMARY, tracked->resreq->tag);
TEST_ASSERT_NOT_NULL(tracked->resreq->u.primary);
TEST_ASSERT_EQUAL_INT(1, tracked->resreq->u.primary->minTasks);
TEST_ASSERT_EQUAL_INT(1, tracked->resreq->u.primary->maxTasks);
}
/* Session-active rejection (Req 5 AC 8 / design.md precondition):
* core_push_events() is only valid between sessions. */
static void
@ -1631,6 +1715,8 @@ int main(void)
RUN_TEST(test_job_rusage_reconciles_mem_reservation_against_usage);
RUN_TEST(test_config_push_decomposes_queues_resources_and_resvals);
RUN_TEST(test_core_build_primary_schedobj_populates_resreq_fp_from_parsed_resreq);
RUN_TEST(test_core_build_job_wires_real_task_count_into_resreq_primary);
RUN_TEST(test_core_build_job_default_task_count_still_one);
RUN_TEST(test_push_events_rejects_with_ebusy_when_session_active_no_mutation);
RUN_TEST(test_push_events_null_core_returns_einval);
RUN_TEST(test_push_events_null_events_with_positive_nevents_returns_einval);