wdgzyjx/myxce
1
0
Fork 0
forked from lei.chen/myxce
Code Pull requests Activity

make unsatisfy_() just print unmeet resources

Browse source
This commit is contained in:
Rongsong Shen 2026-05-13 10:48:38 +00:00
parent febb4a6c4d
commit aae4c672b5
1 changed files with 35 additions and 27 deletions
Download patch file Download diff file Expand all files Collapse all files

62
src/scheduler/src/resource.c
View file

@ -102,9 +102,11 @@ static int unsatisfy_(xce_scheduler_t* sched, int nres, int* idxs, float* total,
count = 0; count = 0;
for (i = 0; i < nres; i++) { for (i = 0; i < nres; i++) {
TRACE("%s resource %d total %f inuse %f uncomit %f wanted %f", __func__,
idxs[i], total[i], inuse[i], uncomitted[i], wanted[i]);
if (total[i] < inuse[i] + uncomitted[i] + wanted[i]) { if (total[i] < inuse[i] + uncomitted[i] + wanted[i]) {
TRACE("%s resource %d total %f inuse %f uncomit %f wanted %f",
__func__, idxs[i], total[i], inuse[i], uncomitted[i],
wanted[i]);
count++; count++;
} }
} }
@ -116,8 +118,8 @@ static int unsatisfy_(xce_scheduler_t* sched, int nres, int* idxs, float* total,
j = 0; j = 0;
for (i = 0; i < nres; i++) { for (i = 0; i < nres; i++) {
if (total[i] < inuse[i] + uncomitted[i] + wanted[i]) { if (total[i] < inuse[i] + uncomitted[i] + wanted[i]) {
xce_resource_t* r = NULL; xce_resource_t* r = NULL;
float need = inuse[i] + uncomitted[i] + wanted[i] - total[i]; float need = inuse[i] + uncomitted[i] + wanted[i] - total[i];
vget(sched->resmgr, idxs[i], (void**)&r); vget(sched->resmgr, idxs[i], (void**)&r);
TRACE("%s unmet resource %d (%s) shortfall %f", __func__, idxs[i], TRACE("%s unmet resource %d (%s) shortfall %f", __func__, idxs[i],
@ -132,14 +134,16 @@ static int unsatisfy_(xce_scheduler_t* sched, int nres, int* idxs, float* total,
return count; return count;
} }
static host_eval_ctx_t* static host_eval_ctx_t* build_host_eval_ctx(const xce_sched_worker_t* w,
build_host_eval_ctx(const xce_sched_worker_t* w, int num_res) int num_res)
{ {
host_eval_ctx_t* ctx = calloc(1, sizeof(*ctx)); host_eval_ctx_t* ctx = calloc(1, sizeof(*ctx));
if (ctx == NULL) return NULL; if (ctx == NULL)
return NULL;
if (num_res <= 0) { if (num_res <= 0) {
/* valid empty context — numvals/strvals remain NULL, bounds checks protect access */ /* valid empty context — numvals/strvals remain NULL, bounds checks
* protect access */
ctx->num_res = 0; ctx->num_res = 0;
return ctx; return ctx;
} }
@ -155,7 +159,8 @@ build_host_eval_ctx(const xce_sched_worker_t* w, int num_res)
for (int i = 0; i < num_res; i++) { for (int i = 0; i < num_res; i++) {
xce_val_t* v = (xce_val_t*)zarray_get((zarray_t*)&w->resvals, i); xce_val_t* v = (xce_val_t*)zarray_get((zarray_t*)&w->resvals, i);
if (v == NULL) continue; if (v == NULL)
continue;
switch (v->which) { switch (v->which) {
case Value_u_fvalue: case Value_u_fvalue:
numvals[i] = (double)v->u.fvalue; numvals[i] = (double)v->u.fvalue;
@ -175,10 +180,10 @@ build_host_eval_ctx(const xce_sched_worker_t* w, int num_res)
return ctx; return ctx;
} }
static void static void free_host_eval_ctx(host_eval_ctx_t* ctx)
free_host_eval_ctx(host_eval_ctx_t* ctx)
{ {
if (ctx == NULL) return; if (ctx == NULL)
return;
free((void*)ctx->numvals); free((void*)ctx->numvals);
free((void*)ctx->strvals); free((void*)ctx->strvals);
free(ctx); free(ctx);
@ -186,7 +191,8 @@ free_host_eval_ctx(host_eval_ctx_t* ctx)
void sched_worker_release_host_ctx(xce_sched_worker_t* w) void sched_worker_release_host_ctx(xce_sched_worker_t* w)
{ {
if (w == NULL) return; if (w == NULL)
return;
free_host_eval_ctx(w->host_eval_ctx); free_host_eval_ctx(w->host_eval_ctx);
w->host_eval_ctx = NULL; w->host_eval_ctx = NULL;
} }
@ -194,11 +200,12 @@ void sched_worker_release_host_ctx(xce_sched_worker_t* w)
int sched_meet_host_selection(xce_scheduler_t* sched, xce_sched_worker_t* w, int sched_meet_host_selection(xce_scheduler_t* sched, xce_sched_worker_t* w,
xce_select_t* s) xce_select_t* s)
{ {
if (s == NULL || s->root == NULL) return 1; if (s == NULL || s->root == NULL)
return 1;
/* Build host context lazily — reused for all jobs against this host. */ /* Build host context lazily — reused for all jobs against this host. */
if (w->host_eval_ctx == NULL) { if (w->host_eval_ctx == NULL) {
int num_res = (int)w->resvals.caps; int num_res = (int)w->resvals.caps;
w->host_eval_ctx = build_host_eval_ctx(w, num_res); w->host_eval_ctx = build_host_eval_ctx(w, num_res);
if (w->host_eval_ctx == NULL) { if (w->host_eval_ctx == NULL) {
/* OOM: conservatively pass host */ /* OOM: conservatively pass host */
@ -216,7 +223,8 @@ int sched_meet_host_selection(xce_scheduler_t* sched, xce_sched_worker_t* w,
return (rc > 0) ? 1 : 0; return (rc > 0) ? 1 : 0;
} }
/* Fallback (no session context — e.g. unit tests or OOM at session start). */ /* Fallback (no session context — e.g. unit tests or OOM at session start).
*/
int rc = eval_select_simple(s->root, w->host_eval_ctx); int rc = eval_select_simple(s->root, w->host_eval_ctx);
return (rc > 0) ? 1 : 0; return (rc > 0) ? 1 : 0;
} }
@ -464,7 +472,7 @@ static int sched_try_alloc_on_host_(xce_sched_worker_t* w, va_list ap)
if (ctx->reason == NULL) { if (ctx->reason == NULL) {
return -errno; return -errno;
} }
add_job_preason(ctx->reason->reasons, PREASON_MATCH_FAIL, w->name, add_job_preason(ctx->reason->reasons, PREASON_MATCH_FAIL, w->name,
NULL); NULL);
@ -496,18 +504,18 @@ static int sched_try_alloc_on_host_(xce_sched_worker_t* w, va_list ap)
int i; int i;
if (ctx) if (ctx)
for (i = 0; i < ctx->unmeet.count; i++) { for (i = 0; i < ctx->unmeet.count; i++) {
xce_resource_t* r = NULL; xce_resource_t* r = NULL;
rc = vget(sched->resmgr, ((int*)ctx->unmeet.idxs.data)[i], rc = vget(sched->resmgr, ((int*)ctx->unmeet.idxs.data)[i],
(void**)&r); (void**)&r);
if (rc < 0) { if (rc < 0) {
continue; continue;
}
add_job_preason(ctx->reason->reasons, PREASON_NO_RESOURCE,
w->name, r->name);
} }
add_job_preason(ctx->reason->reasons, PREASON_NO_RESOURCE, w->name,
r->name);
}
return 0; return 0;
} }