3 #define RRDVAR_MAX_LENGTH 1024
5 static const char *health_default_exec = PLUGINS_DIR "/alarm-notify.sh";
6 static const char *health_default_recipient = "root";
7 int health_enabled = 1;
9 // ----------------------------------------------------------------------------
10 // Health Alarms Log Management
12 static inline void health_alarm_log(RRDHOST *host,
13 uint32_t alarm_id, uint32_t alarm_event_id,
15 const char *name, const char *chart, const char *family,
16 const char *exec, const char *recipient, time_t duration,
17 calculated_number old_value, calculated_number new_value,
18 int old_status, int new_status,
24 debug(D_HEALTH, "Health adding alarm log entry with id: %u", host->health_log.next_log_id);
26 ALARM_ENTRY *ae = callocz(1, sizeof(ALARM_ENTRY));
27 ae->name = strdupz(name);
28 ae->hash_name = simple_hash(ae->name);
31 ae->chart = strdupz(chart);
32 ae->hash_chart = simple_hash(ae->chart);
36 ae->family = strdupz(family);
38 if(exec) ae->exec = strdupz(exec);
39 if(recipient) ae->recipient = strdupz(recipient);
40 if(source) ae->source = strdupz(source);
41 if(units) ae->units = strdupz(units);
42 if(info) ae->info = strdupz(info);
44 ae->unique_id = host->health_log.next_log_id++;
45 ae->alarm_id = alarm_id;
46 ae->alarm_event_id = alarm_event_id;
48 ae->old_value = old_value;
49 ae->new_value = new_value;
50 ae->old_status = old_status;
51 ae->new_status = new_status;
52 ae->duration = duration;
54 ae->delay_up_to_timestamp = when + delay;
56 if(ae->old_status == RRDCALC_STATUS_WARNING || ae->old_status == RRDCALC_STATUS_CRITICAL)
57 ae->non_clear_duration += ae->duration;
60 pthread_rwlock_wrlock(&host->health_log.alarm_log_rwlock);
61 ae->next = host->health_log.alarms;
62 host->health_log.alarms = ae;
63 host->health_log.count++;
64 pthread_rwlock_unlock(&host->health_log.alarm_log_rwlock);
66 // match previous alarms
67 pthread_rwlock_rdlock(&host->health_log.alarm_log_rwlock);
69 for(t = host->health_log.alarms ; t ; t = t->next) {
70 if(t != ae && t->alarm_id == ae->alarm_id) {
71 if(!(t->notifications & HEALTH_ENTRY_NOTIFICATIONS_UPDATED) && !t->updated_by) {
72 t->notifications |= HEALTH_ENTRY_NOTIFICATIONS_UPDATED;
75 if((t->new_status == RRDCALC_STATUS_WARNING || t->new_status == RRDCALC_STATUS_CRITICAL) &&
76 (t->old_status == RRDCALC_STATUS_WARNING || t->old_status == RRDCALC_STATUS_CRITICAL))
77 ae->non_clear_duration += t->non_clear_duration;
80 // no need to continue
85 pthread_rwlock_unlock(&host->health_log.alarm_log_rwlock);
88 // ----------------------------------------------------------------------------
91 static inline int rrdvar_fix_name(char *variable) {
94 if (!isalnum(*variable) && *variable != '.' && *variable != '_') {
105 int rrdvar_compare(void* a, void* b) {
106 if(((RRDVAR *)a)->hash < ((RRDVAR *)b)->hash) return -1;
107 else if(((RRDVAR *)a)->hash > ((RRDVAR *)b)->hash) return 1;
108 else return strcmp(((RRDVAR *)a)->name, ((RRDVAR *)b)->name);
111 static inline RRDVAR *rrdvar_index_add(avl_tree_lock *tree, RRDVAR *rv) {
112 RRDVAR *ret = (RRDVAR *)avl_insert_lock(tree, (avl *)(rv));
114 debug(D_VARIABLES, "Request to insert RRDVAR '%s' into index failed. Already exists.", rv->name);
119 static inline RRDVAR *rrdvar_index_del(avl_tree_lock *tree, RRDVAR *rv) {
120 RRDVAR *ret = (RRDVAR *)avl_remove_lock(tree, (avl *)(rv));
122 error("Request to remove RRDVAR '%s' from index failed. Not Found.", rv->name);
127 static inline RRDVAR *rrdvar_index_find(avl_tree_lock *tree, const char *name, uint32_t hash) {
129 tmp.name = (char *)name;
130 tmp.hash = (hash)?hash:simple_hash(tmp.name);
132 return (RRDVAR *)avl_search_lock(tree, (avl *)&tmp);
135 static inline void rrdvar_free(RRDHOST *host, avl_tree_lock *tree, RRDVAR *rv) {
141 rrdvar_index_del(tree, rv);
147 static inline RRDVAR *rrdvar_create_and_index(const char *scope, avl_tree_lock *tree, const char *name, int type, calculated_number *value) {
148 char *variable = strdupz(name);
149 rrdvar_fix_name(variable);
150 uint32_t hash = simple_hash(variable);
152 RRDVAR *rv = rrdvar_index_find(tree, variable, hash);
154 debug(D_VARIABLES, "Variable '%s' not found in scope '%s'. Creating a new one.", variable, scope);
156 rv = callocz(1, sizeof(RRDVAR));
162 RRDVAR *ret = rrdvar_index_add(tree, rv);
163 if(unlikely(ret != rv)) {
164 debug(D_VARIABLES, "Variable '%s' in scope '%s' already exists", variable, scope);
165 rrdvar_free(NULL, NULL, rv);
169 debug(D_VARIABLES, "Variable '%s' created in scope '%s'", variable, scope);
180 // ----------------------------------------------------------------------------
183 calculated_number rrdvar2number(RRDVAR *rv) {
185 case RRDVAR_TYPE_CALCULATED: {
186 calculated_number *n = (calculated_number *)rv->value;
190 case RRDVAR_TYPE_TIME_T: {
191 time_t *n = (time_t *)rv->value;
195 case RRDVAR_TYPE_COLLECTED: {
196 collected_number *n = (collected_number *)rv->value;
200 case RRDVAR_TYPE_TOTAL: {
201 total_number *n = (total_number *)rv->value;
205 case RRDVAR_TYPE_INT: {
206 int *n = (int *)rv->value;
211 error("I don't know how to convert RRDVAR type %d to calculated_number", rv->type);
216 void dump_variable(void *data) {
217 RRDVAR *rv = (RRDVAR *)data;
218 debug(D_HEALTH, "%50s : %20.5Lf", rv->name, rrdvar2number(rv));
221 int health_variable_lookup(const char *variable, uint32_t hash, RRDCALC *rc, calculated_number *result) {
222 RRDSET *st = rc->rrdset;
227 rv = rrdvar_index_find(&st->variables_root_index, variable, hash);
229 *result = rrdvar2number(rv);
233 rv = rrdvar_index_find(&st->rrdfamily->variables_root_index, variable, hash);
235 *result = rrdvar2number(rv);
239 rv = rrdvar_index_find(&st->rrdhost->variables_root_index, variable, hash);
241 *result = rrdvar2number(rv);
245 debug(D_HEALTH, "Available local chart '%s' variables:", st->id);
246 avl_traverse_lock(&st->variables_root_index, dump_variable);
248 debug(D_HEALTH, "Available family '%s' variables:", st->rrdfamily->family);
249 avl_traverse_lock(&st->rrdfamily->variables_root_index, dump_variable);
251 debug(D_HEALTH, "Available host '%s' variables:", st->rrdhost->hostname);
252 avl_traverse_lock(&st->rrdhost->variables_root_index, dump_variable);
257 // ----------------------------------------------------------------------------
258 // RRDSETVAR management
260 RRDSETVAR *rrdsetvar_create(RRDSET *st, const char *variable, int type, void *value, uint32_t options) {
261 debug(D_VARIABLES, "RRDVARSET create for chart id '%s' name '%s' with variable name '%s'", st->id, st->name, variable);
262 RRDSETVAR *rs = (RRDSETVAR *)callocz(1, sizeof(RRDSETVAR));
264 char buffer[RRDVAR_MAX_LENGTH + 1];
265 snprintfz(buffer, RRDVAR_MAX_LENGTH, "%s.%s", st->id, variable);
266 rs->fullid = strdupz(buffer);
268 snprintfz(buffer, RRDVAR_MAX_LENGTH, "%s.%s", st->name, variable);
269 rs->fullname = strdupz(buffer);
271 rs->variable = strdupz(variable);
275 rs->options = options;
278 rs->local = rrdvar_create_and_index("local", &st->variables_root_index, rs->variable, rs->type, rs->value);
279 rs->family = rrdvar_create_and_index("family", &st->rrdfamily->variables_root_index, rs->fullid, rs->type, rs->value);
280 rs->host = rrdvar_create_and_index("host", &st->rrdhost->variables_root_index, rs->fullid, rs->type, rs->value);
281 rs->family_name = rrdvar_create_and_index("family", &st->rrdfamily->variables_root_index, rs->fullname, rs->type, rs->value);
282 rs->host_name = rrdvar_create_and_index("host", &st->rrdhost->variables_root_index, rs->fullname, rs->type, rs->value);
284 rs->next = st->variables;
290 void rrdsetvar_rename_all(RRDSET *st) {
291 debug(D_VARIABLES, "RRDSETVAR rename for chart id '%s' name '%s'", st->id, st->name);
293 // only these 2 can change name
297 char buffer[RRDVAR_MAX_LENGTH + 1];
298 RRDSETVAR *rs, *next = st->variables;
302 snprintfz(buffer, RRDVAR_MAX_LENGTH, "%s.%s", st->name, rs->variable);
304 if (strcmp(buffer, rs->fullname)) {
306 rrdvar_free(st->rrdhost, &st->rrdfamily->variables_root_index, rs->family_name);
307 rrdvar_free(st->rrdhost, &st->rrdhost->variables_root_index, rs->host_name);
310 rs->fullname = strdupz(st->name);
311 rs->family_name = rrdvar_create_and_index("family", &st->rrdfamily->variables_root_index, rs->fullname, rs->type, rs->value);
312 rs->host_name = rrdvar_create_and_index("host", &st->rrdhost->variables_root_index, rs->fullname, rs->type, rs->value);
316 rrdsetcalc_link_matching(st);
319 void rrdsetvar_free(RRDSETVAR *rs) {
320 RRDSET *st = rs->rrdset;
321 debug(D_VARIABLES, "RRDSETVAR free for chart id '%s' name '%s', variable '%s'", st->id, st->name, rs->variable);
323 rrdvar_free(st->rrdhost, &st->variables_root_index, rs->local);
324 rrdvar_free(st->rrdhost, &st->rrdfamily->variables_root_index, rs->family);
325 rrdvar_free(st->rrdhost, &st->rrdhost->variables_root_index, rs->host);
326 rrdvar_free(st->rrdhost, &st->rrdfamily->variables_root_index, rs->family_name);
327 rrdvar_free(st->rrdhost, &st->rrdhost->variables_root_index, rs->host_name);
329 if(st->variables == rs) {
330 st->variables = rs->next;
334 for (t = st->variables; t && t->next != rs; t = t->next);
335 if(!t) error("RRDSETVAR '%s' not found in chart '%s' variables linked list", rs->fullname, st->id);
336 else t->next = rs->next;
345 // ----------------------------------------------------------------------------
346 // RRDDIMVAR management
348 #define RRDDIMVAR_ID_MAX 1024
350 RRDDIMVAR *rrddimvar_create(RRDDIM *rd, int type, const char *prefix, const char *suffix, void *value, uint32_t options) {
351 RRDSET *st = rd->rrdset;
353 debug(D_VARIABLES, "RRDDIMSET create for chart id '%s' name '%s', dimension id '%s', name '%s%s%s'", st->id, st->name, rd->id, (prefix)?prefix:"", rd->name, (suffix)?suffix:"");
355 if(!prefix) prefix = "";
356 if(!suffix) suffix = "";
358 char buffer[RRDDIMVAR_ID_MAX + 1];
359 RRDDIMVAR *rs = (RRDDIMVAR *)callocz(1, sizeof(RRDDIMVAR));
361 rs->prefix = strdupz(prefix);
362 rs->suffix = strdupz(suffix);
364 snprintfz(buffer, RRDDIMVAR_ID_MAX, "%s%s%s", rs->prefix, rd->id, rs->suffix);
365 rs->id = strdupz(buffer);
367 snprintfz(buffer, RRDDIMVAR_ID_MAX, "%s%s%s", rs->prefix, rd->name, rs->suffix);
368 rs->name = strdupz(buffer);
370 snprintfz(buffer, RRDDIMVAR_ID_MAX, "%s.%s", rd->rrdset->id, rs->id);
371 rs->fullidid = strdupz(buffer);
373 snprintfz(buffer, RRDDIMVAR_ID_MAX, "%s.%s", rd->rrdset->id, rs->name);
374 rs->fullidname = strdupz(buffer);
376 snprintfz(buffer, RRDDIMVAR_ID_MAX, "%s.%s", rd->rrdset->name, rs->id);
377 rs->fullnameid = strdupz(buffer);
379 snprintfz(buffer, RRDDIMVAR_ID_MAX, "%s.%s", rd->rrdset->name, rs->name);
380 rs->fullnamename = strdupz(buffer);
384 rs->options = options;
387 rs->local_id = rrdvar_create_and_index("local", &st->variables_root_index, rs->id, rs->type, rs->value);
388 rs->local_name = rrdvar_create_and_index("local", &st->variables_root_index, rs->name, rs->type, rs->value);
390 rs->family_id = rrdvar_create_and_index("family", &st->rrdfamily->variables_root_index, rs->id, rs->type, rs->value);
391 rs->family_name = rrdvar_create_and_index("family", &st->rrdfamily->variables_root_index, rs->name, rs->type, rs->value);
393 rs->host_fullidid = rrdvar_create_and_index("host", &st->rrdhost->variables_root_index, rs->fullidid, rs->type, rs->value);
394 rs->host_fullidname = rrdvar_create_and_index("host", &st->rrdhost->variables_root_index, rs->fullidname, rs->type, rs->value);
395 rs->host_fullnameid = rrdvar_create_and_index("host", &st->rrdhost->variables_root_index, rs->fullnameid, rs->type, rs->value);
396 rs->host_fullnamename = rrdvar_create_and_index("host", &st->rrdhost->variables_root_index, rs->fullnamename, rs->type, rs->value);
398 rs->next = rd->variables;
404 void rrddimvar_rename_all(RRDDIM *rd) {
405 RRDSET *st = rd->rrdset;
406 debug(D_VARIABLES, "RRDDIMSET rename for chart id '%s' name '%s', dimension id '%s', name '%s'", st->id, st->name, rd->id, rd->name);
408 RRDDIMVAR *rs, *next = rd->variables;
412 if (strcmp(rd->name, rs->name)) {
413 char buffer[RRDDIMVAR_ID_MAX + 1];
417 rrdvar_free(st->rrdhost, &st->variables_root_index, rs->local_name);
419 snprintfz(buffer, RRDDIMVAR_ID_MAX, "%s%s%s", rs->prefix, rd->name, rs->suffix);
420 rs->name = strdupz(buffer);
421 rs->local_name = rrdvar_create_and_index("local", &st->variables_root_index, rs->name, rs->type, rs->value);
423 rrdvar_free(st->rrdhost, &st->rrdhost->variables_root_index, rs->host_fullidname);
424 freez(rs->fullidname);
425 snprintfz(buffer, RRDDIMVAR_ID_MAX, "%s.%s", st->id, rs->name);
426 rs->fullidname = strdupz(buffer);
427 rs->host_fullidname = rrdvar_create_and_index("host", &st->rrdhost->variables_root_index,
428 rs->fullidname, rs->type, rs->value);
431 rrdvar_free(st->rrdhost, &st->rrdhost->variables_root_index, rs->host_fullnameid);
432 freez(rs->fullnameid);
433 snprintfz(buffer, RRDDIMVAR_ID_MAX, "%s.%s", st->name, rs->id);
434 rs->fullnameid = strdupz(buffer);
435 rs->host_fullnameid = rrdvar_create_and_index("host", &st->rrdhost->variables_root_index,
436 rs->fullnameid, rs->type, rs->value);
439 rrdvar_free(st->rrdhost, &st->rrdhost->variables_root_index, rs->host_fullnamename);
440 freez(rs->fullnamename);
441 snprintfz(buffer, RRDDIMVAR_ID_MAX, "%s.%s", st->name, rs->name);
442 rs->fullnamename = strdupz(buffer);
443 rs->host_fullnamename = rrdvar_create_and_index("host", &st->rrdhost->variables_root_index,
444 rs->fullnamename, rs->type, rs->value);
449 void rrddimvar_free(RRDDIMVAR *rs) {
450 RRDDIM *rd = rs->rrddim;
451 RRDSET *st = rd->rrdset;
452 debug(D_VARIABLES, "RRDDIMSET free for chart id '%s' name '%s', dimension id '%s', name '%s', prefix='%s', suffix='%s'", st->id, st->name, rd->id, rd->name, rs->prefix, rs->suffix);
454 rrdvar_free(st->rrdhost, &st->variables_root_index, rs->local_id);
455 rrdvar_free(st->rrdhost, &st->variables_root_index, rs->local_name);
457 rrdvar_free(st->rrdhost, &st->rrdfamily->variables_root_index, rs->family_id);
458 rrdvar_free(st->rrdhost, &st->rrdfamily->variables_root_index, rs->family_name);
460 rrdvar_free(st->rrdhost, &st->rrdhost->variables_root_index, rs->host_fullidid);
461 rrdvar_free(st->rrdhost, &st->rrdhost->variables_root_index, rs->host_fullidname);
462 rrdvar_free(st->rrdhost, &st->rrdhost->variables_root_index, rs->host_fullnameid);
463 rrdvar_free(st->rrdhost, &st->rrdhost->variables_root_index, rs->host_fullnamename);
465 if(rd->variables == rs) {
466 debug(D_VARIABLES, "RRDDIMSET removing first entry for chart id '%s' name '%s', dimension id '%s', name '%s'", st->id, st->name, rd->id, rd->name);
467 rd->variables = rs->next;
470 debug(D_VARIABLES, "RRDDIMSET removing non-first entry for chart id '%s' name '%s', dimension id '%s', name '%s'", st->id, st->name, rd->id, rd->name);
472 for (t = rd->variables; t && t->next != rs; t = t->next) ;
473 if(!t) error("RRDDIMVAR '%s' not found in dimension '%s/%s' variables linked list", rs->name, st->id, rd->id);
474 else t->next = rs->next;
482 freez(rs->fullidname);
483 freez(rs->fullnameid);
484 freez(rs->fullnamename);
488 // ----------------------------------------------------------------------------
489 // RRDCALC management
491 static inline const char *rrdcalc_status2string(int status) {
493 case RRDCALC_STATUS_REMOVED:
496 case RRDCALC_STATUS_UNDEFINED:
499 case RRDCALC_STATUS_UNINITIALIZED:
500 return "UNINITIALIZED";
502 case RRDCALC_STATUS_CLEAR:
505 case RRDCALC_STATUS_RAISED:
508 case RRDCALC_STATUS_WARNING:
511 case RRDCALC_STATUS_CRITICAL:
515 error("Unknown alarm status %d", status);
520 static void rrdsetcalc_link(RRDSET *st, RRDCALC *rc) {
521 debug(D_HEALTH, "Health linking alarm '%s.%s' to chart '%s' of host '%s'", rc->chart?rc->chart:"NOCHART", rc->name, st->id, st->rrdhost->hostname);
523 rc->last_status_change = time(NULL);
526 rc->rrdset_next = st->alarms;
527 rc->rrdset_prev = NULL;
530 rc->rrdset_next->rrdset_prev = rc;
534 if(rc->update_every < rc->rrdset->update_every) {
535 error("Health alarm '%s.%s' has update every %d, less than chart update every %d. Setting alarm update frequency to %d.", rc->rrdset->id, rc->name, rc->update_every, rc->rrdset->update_every, rc->rrdset->update_every);
536 rc->update_every = rc->rrdset->update_every;
539 if(!isnan(rc->green) && isnan(st->green)) {
540 debug(D_HEALTH, "Health alarm '%s.%s' green threshold set from %Lf to %Lf.", rc->rrdset->id, rc->name, rc->rrdset->green, rc->green);
541 st->green = rc->green;
544 if(!isnan(rc->red) && isnan(st->red)) {
545 debug(D_HEALTH, "Health alarm '%s.%s' red threshold set from %Lf to %Lf.", rc->rrdset->id, rc->name, rc->rrdset->red, rc->red);
549 rc->local = rrdvar_create_and_index("local", &st->variables_root_index, rc->name, RRDVAR_TYPE_CALCULATED, &rc->value);
550 rc->family = rrdvar_create_and_index("family", &st->rrdfamily->variables_root_index, rc->name, RRDVAR_TYPE_CALCULATED, &rc->value);
552 char fullname[RRDVAR_MAX_LENGTH + 1];
553 snprintfz(fullname, RRDVAR_MAX_LENGTH, "%s.%s", st->id, rc->name);
554 rc->hostid = rrdvar_create_and_index("host", &st->rrdhost->variables_root_index, fullname, RRDVAR_TYPE_CALCULATED, &rc->value);
556 snprintfz(fullname, RRDVAR_MAX_LENGTH, "%s.%s", st->name, rc->name);
557 rc->hostname = rrdvar_create_and_index("host", &st->rrdhost->variables_root_index, fullname, RRDVAR_TYPE_CALCULATED, &rc->value);
559 if(!rc->units) rc->units = strdupz(st->units);
562 time_t now = time(NULL);
563 health_alarm_log(st->rrdhost, rc->id, rc->next_event_id++, now, rc->name, rc->rrdset->id, rc->rrdset->family, rc->exec, rc->recipient, now - rc->last_status_change, rc->old_value, rc->value, rc->status, RRDCALC_STATUS_UNINITIALIZED, rc->source, rc->units, rc->info, 0);
567 static inline int rrdcalc_is_matching_this_rrdset(RRDCALC *rc, RRDSET *st) {
568 if( (rc->hash_chart == st->hash && !strcmp(rc->chart, st->id)) ||
569 (rc->hash_chart == st->hash_name && !strcmp(rc->chart, st->name)))
575 // this has to be called while the RRDHOST is locked
576 inline void rrdsetcalc_link_matching(RRDSET *st) {
577 // debug(D_HEALTH, "find matching alarms for chart '%s'", st->id);
580 for(rc = st->rrdhost->alarms; rc ; rc = rc->next) {
581 if(unlikely(rc->rrdset))
584 if(unlikely(rrdcalc_is_matching_this_rrdset(rc, st)))
585 rrdsetcalc_link(st, rc);
589 // this has to be called while the RRDHOST is locked
590 inline void rrdsetcalc_unlink(RRDCALC *rc) {
591 RRDSET *st = rc->rrdset;
594 debug(D_HEALTH, "Requested to unlink RRDCALC '%s.%s' which is not linked to any RRDSET", rc->chart?rc->chart:"NOCHART", rc->name);
595 error("Requested to unlink RRDCALC '%s.%s' which is not linked to any RRDSET", rc->chart?rc->chart:"NOCHART", rc->name);
600 time_t now = time(NULL);
601 health_alarm_log(st->rrdhost, rc->id, rc->next_event_id++, now, rc->name, rc->rrdset->id, rc->rrdset->family, rc->exec, rc->recipient, now - rc->last_status_change, rc->old_value, rc->value, rc->status, RRDCALC_STATUS_REMOVED, rc->source, rc->units, rc->info, 0);
604 RRDHOST *host = st->rrdhost;
606 debug(D_HEALTH, "Health unlinking alarm '%s.%s' from chart '%s' of host '%s'", rc->chart?rc->chart:"NOCHART", rc->name, st->id, host->hostname);
610 rc->rrdset_prev->rrdset_next = rc->rrdset_next;
613 rc->rrdset_next->rrdset_prev = rc->rrdset_prev;
616 st->alarms = rc->rrdset_next;
618 rc->rrdset_prev = rc->rrdset_next = NULL;
620 rrdvar_free(st->rrdhost, &st->variables_root_index, rc->local);
623 rrdvar_free(st->rrdhost, &st->rrdfamily->variables_root_index, rc->family);
626 rrdvar_free(st->rrdhost, &st->rrdhost->variables_root_index, rc->hostid);
629 rrdvar_free(st->rrdhost, &st->rrdhost->variables_root_index, rc->hostname);
634 // RRDCALC will remain in RRDHOST
635 // so that if the matching chart is found in the future
636 // it will be applied automatically
639 RRDCALC *rrdcalc_find(RRDSET *st, const char *name) {
641 uint32_t hash = simple_hash(name);
643 for( rc = st->alarms; rc ; rc = rc->rrdset_next ) {
644 if(unlikely(rc->hash == hash && !strcmp(rc->name, name)))
651 static inline int rrdcalc_exists(RRDHOST *host, const char *chart, const char *name, uint32_t hash_chart, uint32_t hash_name) {
654 if(unlikely(!chart)) {
655 error("attempt to find RRDCALC '%s' without giving a chart name", name);
659 if(unlikely(!hash_chart)) hash_chart = simple_hash(chart);
660 if(unlikely(!hash_name)) hash_name = simple_hash(name);
662 // make sure it does not already exist
663 for(rc = host->alarms; rc ; rc = rc->next) {
664 if (unlikely(rc->chart && rc->hash == hash_name && rc->hash_chart == hash_chart && !strcmp(name, rc->name) && !strcmp(chart, rc->chart))) {
665 debug(D_HEALTH, "Health alarm '%s.%s' already exists in host '%s'.", chart, name, host->hostname);
666 error("Health alarm '%s.%s' already exists in host '%s'.", chart, name, host->hostname);
674 static inline uint32_t rrdcalc_get_unique_id(RRDHOST *host, const char *chart, const char *name, uint32_t *next_event_id) {
676 uint32_t hash_chart = simple_hash(chart);
677 uint32_t hash_name = simple_hash(name);
679 // re-use old IDs, by looking them up in the alarm log
681 for(ae = host->health_log.alarms; ae ;ae = ae->next) {
682 if(unlikely(ae->hash_name == hash_name && ae->hash_chart == hash_chart && !strcmp(name, ae->name) && !strcmp(chart, ae->chart))) {
683 if(next_event_id) *next_event_id = ae->alarm_event_id + 1;
689 return host->health_log.next_alarm_id++;
692 static inline void rrdcalc_create_part2(RRDHOST *host, RRDCALC *rc) {
693 rrdhost_check_rdlock(host);
695 if(rc->calculation) {
696 rc->calculation->status = &rc->status;
697 rc->calculation->this = &rc->value;
698 rc->calculation->after = &rc->db_after;
699 rc->calculation->before = &rc->db_before;
700 rc->calculation->rrdcalc = rc;
704 rc->warning->status = &rc->status;
705 rc->warning->this = &rc->value;
706 rc->warning->after = &rc->db_after;
707 rc->warning->before = &rc->db_before;
708 rc->warning->rrdcalc = rc;
712 rc->critical->status = &rc->status;
713 rc->critical->this = &rc->value;
714 rc->critical->after = &rc->db_after;
715 rc->critical->before = &rc->db_before;
716 rc->critical->rrdcalc = rc;
719 // link it to the host
720 if(likely(host->alarms)) {
723 for(t = host->alarms; t && t->next ; t = t->next) ;
730 // link it to its chart
732 for(st = host->rrdset_root; st ; st = st->next) {
733 if(rrdcalc_is_matching_this_rrdset(rc, st)) {
734 rrdsetcalc_link(st, rc);
740 static inline RRDCALC *rrdcalc_create(RRDHOST *host, RRDCALCTEMPLATE *rt, const char *chart) {
742 debug(D_HEALTH, "Health creating dynamic alarm (from template) '%s.%s'", chart, rt->name);
744 if(rrdcalc_exists(host, chart, rt->name, 0, 0))
747 RRDCALC *rc = callocz(1, sizeof(RRDCALC));
748 rc->next_event_id = 1;
749 rc->id = rrdcalc_get_unique_id(host, chart, rt->name, &rc->next_event_id);
750 rc->name = strdupz(rt->name);
751 rc->hash = simple_hash(rc->name);
752 rc->chart = strdupz(chart);
753 rc->hash_chart = simple_hash(rc->chart);
755 if(rt->dimensions) rc->dimensions = strdupz(rt->dimensions);
757 rc->green = rt->green;
762 rc->delay_up_duration = rt->delay_up_duration;
763 rc->delay_down_duration = rt->delay_down_duration;
764 rc->delay_max_duration = rt->delay_max_duration;
765 rc->delay_multiplier = rt->delay_multiplier;
767 rc->group = rt->group;
768 rc->after = rt->after;
769 rc->before = rt->before;
770 rc->update_every = rt->update_every;
771 rc->options = rt->options;
773 if(rt->exec) rc->exec = strdupz(rt->exec);
774 if(rt->recipient) rc->recipient = strdupz(rt->recipient);
775 if(rt->source) rc->source = strdupz(rt->source);
776 if(rt->units) rc->units = strdupz(rt->units);
777 if(rt->info) rc->info = strdupz(rt->info);
779 if(rt->calculation) {
780 rc->calculation = expression_parse(rt->calculation->source, NULL, NULL);
782 error("Health alarm '%s.%s': failed to parse calculation expression '%s'", chart, rt->name, rt->calculation->source);
785 rc->warning = expression_parse(rt->warning->source, NULL, NULL);
787 error("Health alarm '%s.%s': failed to re-parse warning expression '%s'", chart, rt->name, rt->warning->source);
790 rc->critical = expression_parse(rt->critical->source, NULL, NULL);
792 error("Health alarm '%s.%s': failed to re-parse critical expression '%s'", chart, rt->name, rt->critical->source);
795 debug(D_HEALTH, "Health runtime added alarm '%s.%s': exec '%s', recipient '%s', green %Lf, red %Lf, lookup: group %d, after %d, before %d, options %u, dimensions '%s', update every %d, calculation '%s', warning '%s', critical '%s', source '%s', delay up %d, delay down %d, delay max %d, delay_multiplier %f",
796 (rc->chart)?rc->chart:"NOCHART",
798 (rc->exec)?rc->exec:"DEFAULT",
799 (rc->recipient)?rc->recipient:"DEFAULT",
806 (rc->dimensions)?rc->dimensions:"NONE",
808 (rc->calculation)?rc->calculation->parsed_as:"NONE",
809 (rc->warning)?rc->warning->parsed_as:"NONE",
810 (rc->critical)?rc->critical->parsed_as:"NONE",
812 rc->delay_up_duration,
813 rc->delay_down_duration,
814 rc->delay_max_duration,
818 rrdcalc_create_part2(host, rc);
822 void rrdcalc_free(RRDHOST *host, RRDCALC *rc) {
825 debug(D_HEALTH, "Health removing alarm '%s.%s' of host '%s'", rc->chart?rc->chart:"NOCHART", rc->name, host->hostname);
827 // unlink it from RRDSET
828 if(rc->rrdset) rrdsetcalc_unlink(rc);
830 // unlink it from RRDHOST
831 if(unlikely(rc == host->alarms))
832 host->alarms = rc->next;
834 else if(likely(host->alarms)) {
835 RRDCALC *t, *last = host->alarms;
836 for(t = last->next; t && t != rc; last = t, t = t->next) ;
838 last->next = rc->next;
840 error("Cannot unlink alarm '%s.%s' from host '%s': not found", rc->chart?rc->chart:"NOCHART", rc->name, host->hostname);
843 error("Cannot unlink unlink '%s.%s' from host '%s': This host does not have any calculations", rc->chart?rc->chart:"NOCHART", rc->name, host->hostname);
845 expression_free(rc->calculation);
846 expression_free(rc->warning);
847 expression_free(rc->critical);
852 freez(rc->dimensions);
854 freez(rc->recipient);
861 // ----------------------------------------------------------------------------
862 // RRDCALCTEMPLATE management
864 void rrdcalctemplate_link_matching(RRDSET *st) {
867 for(rt = st->rrdhost->templates; rt ; rt = rt->next) {
868 if(rt->hash_context == st->hash_context && !strcmp(rt->context, st->context)) {
869 RRDCALC *rc = rrdcalc_create(st->rrdhost, rt, st->id);
871 error("Health tried to create alarm from template '%s', but it failed", rt->name);
873 #ifdef NETDATA_INTERNAL_CHECKS
874 else if(rc->rrdset != st)
875 error("Health alarm '%s.%s' should be linked to chart '%s', but it is not", rc->chart?rc->chart:"NOCHART", rc->name, st->id);
881 static inline void rrdcalctemplate_free(RRDHOST *host, RRDCALCTEMPLATE *rt) {
882 debug(D_HEALTH, "Health removing template '%s' of host '%s'", rt->name, host->hostname);
884 if(host->templates) {
885 if(host->templates == rt) {
886 host->templates = rt->next;
889 RRDCALCTEMPLATE *t, *last = host->templates;
890 for (t = last->next; t && t != rt; last = t, t = t->next ) ;
891 if(last && last->next == rt) {
892 last->next = rt->next;
896 error("Cannot find RRDCALCTEMPLATE '%s' linked in host '%s'", rt->name, host->hostname);
900 expression_free(rt->calculation);
901 expression_free(rt->warning);
902 expression_free(rt->critical);
906 freez(rt->recipient);
911 freez(rt->dimensions);
915 // ----------------------------------------------------------------------------
916 // load health configuration
918 #define HEALTH_CONF_MAX_LINE 4096
920 #define HEALTH_ALARM_KEY "alarm"
921 #define HEALTH_TEMPLATE_KEY "template"
922 #define HEALTH_ON_KEY "on"
923 #define HEALTH_LOOKUP_KEY "lookup"
924 #define HEALTH_CALC_KEY "calc"
925 #define HEALTH_EVERY_KEY "every"
926 #define HEALTH_GREEN_KEY "green"
927 #define HEALTH_RED_KEY "red"
928 #define HEALTH_WARN_KEY "warn"
929 #define HEALTH_CRIT_KEY "crit"
930 #define HEALTH_EXEC_KEY "exec"
931 #define HEALTH_RECIPIENT_KEY "to"
932 #define HEALTH_UNITS_KEY "units"
933 #define HEALTH_INFO_KEY "info"
934 #define HEALTH_DELAY_KEY "delay"
936 static inline int rrdcalc_add_alarm_from_config(RRDHOST *host, RRDCALC *rc) {
938 error("Health configuration for alarm '%s' does not have a chart", rc->name);
942 if(!rc->update_every) {
943 error("Health configuration for alarm '%s.%s' has no frequency (parameter 'every'). Ignoring it.", rc->chart?rc->chart:"NOCHART", rc->name);
947 if(!RRDCALC_HAS_DB_LOOKUP(rc) && !rc->warning && !rc->critical) {
948 error("Health configuration for alarm '%s.%s' is useless (no calculation, no warning and no critical evaluation)", rc->chart?rc->chart:"NOCHART", rc->name);
952 if (rrdcalc_exists(host, rc->chart, rc->name, rc->hash_chart, rc->hash))
955 rc->id = rrdcalc_get_unique_id(&localhost, rc->chart, rc->name, &rc->next_event_id);
957 debug(D_HEALTH, "Health configuration adding alarm '%s.%s' (%u): exec '%s', recipient '%s', green %Lf, red %Lf, lookup: group %d, after %d, before %d, options %u, dimensions '%s', update every %d, calculation '%s', warning '%s', critical '%s', source '%s', delay up %d, delay down %d, delay max %d, delay_multiplier %f",
958 rc->chart?rc->chart:"NOCHART",
961 (rc->exec)?rc->exec:"DEFAULT",
962 (rc->recipient)?rc->recipient:"DEFAULT",
969 (rc->dimensions)?rc->dimensions:"NONE",
971 (rc->calculation)?rc->calculation->parsed_as:"NONE",
972 (rc->warning)?rc->warning->parsed_as:"NONE",
973 (rc->critical)?rc->critical->parsed_as:"NONE",
975 rc->delay_up_duration,
976 rc->delay_down_duration,
977 rc->delay_max_duration,
981 rrdcalc_create_part2(host, rc);
985 static inline int rrdcalctemplate_add_template_from_config(RRDHOST *host, RRDCALCTEMPLATE *rt) {
986 if(unlikely(!rt->context)) {
987 error("Health configuration for template '%s' does not have a context", rt->name);
991 if(unlikely(!rt->update_every)) {
992 error("Health configuration for template '%s' has no frequency (parameter 'every'). Ignoring it.", rt->name);
996 if(unlikely(!RRDCALCTEMPLATE_HAS_CALCULATION(rt) && !rt->warning && !rt->critical)) {
997 error("Health configuration for template '%s' is useless (no calculation, no warning and no critical evaluation)", rt->name);
1001 RRDCALCTEMPLATE *t, *last = NULL;
1002 for (t = host->templates; t ; last = t, t = t->next) {
1003 if(unlikely(t->hash_name == rt->hash_name && !strcmp(t->name, rt->name))) {
1004 error("Health configuration template '%s' already exists for host '%s'.", rt->name, host->hostname);
1009 debug(D_HEALTH, "Health configuration adding template '%s': context '%s', exec '%s', recipient '%s', green %Lf, red %Lf, lookup: group %d, after %d, before %d, options %u, dimensions '%s', update every %d, calculation '%s', warning '%s', critical '%s', source '%s', delay up %d, delay down %d, delay max %d, delay_multiplier %f",
1011 (rt->context)?rt->context:"NONE",
1012 (rt->exec)?rt->exec:"DEFAULT",
1013 (rt->recipient)?rt->recipient:"DEFAULT",
1020 (rt->dimensions)?rt->dimensions:"NONE",
1022 (rt->calculation)?rt->calculation->parsed_as:"NONE",
1023 (rt->warning)?rt->warning->parsed_as:"NONE",
1024 (rt->critical)?rt->critical->parsed_as:"NONE",
1026 rt->delay_up_duration,
1027 rt->delay_down_duration,
1028 rt->delay_max_duration,
1029 rt->delay_multiplier
1036 rt->next = host->templates;
1037 host->templates = rt;
1043 static inline int health_parse_duration(char *string, int *result) {
1044 // make sure it is a number
1045 if(!*string || !(isdigit(*string) || *string == '+' || *string == '-')) {
1051 calculated_number n = strtold(string, &e);
1055 *result = (int) (n * 86400 * 365);
1058 *result = (int) (n * 86400 * 30);
1061 *result = (int) (n * 86400 * 7);
1064 *result = (int) (n * 86400);
1067 *result = (int) (n * 3600);
1070 *result = (int) (n * 60);
1075 *result = (int) (n);
1085 static inline int health_parse_delay(
1086 size_t line, const char *path, const char *file, char *string,
1087 int *delay_up_duration,
1088 int *delay_down_duration,
1089 int *delay_max_duration,
1090 float *delay_multiplier) {
1093 char given_down = 0;
1095 char given_multiplier = 0;
1101 while(*s && !isspace(*s)) s++;
1102 while(*s && isspace(*s)) *s++ = '\0';
1107 while(*s && !isspace(*s)) s++;
1108 while(*s && isspace(*s)) *s++ = '\0';
1110 if(!strcasecmp(key, "up")) {
1111 if (!health_parse_duration(value, delay_up_duration)) {
1112 error("Health configuration at line %zu of file '%s/%s': invalid value '%s' for '%s' keyword",
1113 line, path, file, value, key);
1117 else if(!strcasecmp(key, "down")) {
1118 if (!health_parse_duration(value, delay_down_duration)) {
1119 error("Health configuration at line %zu of file '%s/%s': invalid value '%s' for '%s' keyword",
1120 line, path, file, value, key);
1122 else given_down = 1;
1124 else if(!strcasecmp(key, "multiplier")) {
1125 *delay_multiplier = strtof(value, NULL);
1126 if(isnan(*delay_multiplier) || isinf(*delay_multiplier) || islessequal(*delay_multiplier, 0)) {
1127 error("Health configuration at line %zu of file '%s/%s': invalid value '%s' for '%s' keyword",
1128 line, path, file, value, key);
1130 else given_multiplier = 1;
1132 else if(!strcasecmp(key, "max")) {
1133 if (!health_parse_duration(value, delay_max_duration)) {
1134 error("Health configuration at line %zu of file '%s/%s': invalid value '%s' for '%s' keyword",
1135 line, path, file, value, key);
1140 error("Health configuration at line %zu of file '%s/%s': unknown keyword '%s'",
1141 line, path, file, key);
1146 *delay_up_duration = 0;
1149 *delay_down_duration = 0;
1151 if(!given_multiplier)
1152 *delay_multiplier = 1.0;
1155 if((*delay_max_duration) < (*delay_up_duration) * (*delay_multiplier))
1156 *delay_max_duration = (*delay_up_duration) * (*delay_multiplier);
1158 if((*delay_max_duration) < (*delay_down_duration) * (*delay_multiplier))
1159 *delay_max_duration = (*delay_down_duration) * (*delay_multiplier);
1165 static inline int health_parse_db_lookup(
1166 size_t line, const char *path, const char *file, char *string,
1167 int *group_method, int *after, int *before, int *every,
1168 uint32_t *options, char **dimensions
1170 debug(D_HEALTH, "Health configuration parsing database lookup %zu@%s/%s: %s", line, path, file, string);
1172 if(*dimensions) freez(*dimensions);
1179 char *s = string, *key;
1181 // first is the group method
1183 while(*s && !isspace(*s)) s++;
1184 while(*s && isspace(*s)) *s++ = '\0';
1186 error("Health configuration invalid chart calculation at line %zu of file '%s/%s': expected group method followed by the 'after' time, but got '%s'",
1187 line, path, file, key);
1191 if((*group_method = web_client_api_request_v1_data_group(key, -1)) == -1) {
1192 error("Health configuration at line %zu of file '%s/%s': invalid group method '%s'",
1193 line, path, file, key);
1197 // then is the 'after' time
1199 while(*s && !isspace(*s)) s++;
1200 while(*s && isspace(*s)) *s++ = '\0';
1202 if(!health_parse_duration(key, after)) {
1203 error("Health configuration at line %zu of file '%s/%s': invalid duration '%s' after group method",
1204 line, path, file, key);
1209 *every = abs(*after);
1211 // now we may have optional parameters
1214 while(*s && !isspace(*s)) s++;
1215 while(*s && isspace(*s)) *s++ = '\0';
1218 if(!strcasecmp(key, "at")) {
1220 while(*s && !isspace(*s)) s++;
1221 while(*s && isspace(*s)) *s++ = '\0';
1223 if (!health_parse_duration(value, before)) {
1224 error("Health configuration at line %zu of file '%s/%s': invalid duration '%s' for '%s' keyword",
1225 line, path, file, value, key);
1228 else if(!strcasecmp(key, HEALTH_EVERY_KEY)) {
1230 while(*s && !isspace(*s)) s++;
1231 while(*s && isspace(*s)) *s++ = '\0';
1233 if (!health_parse_duration(value, every)) {
1234 error("Health configuration at line %zu of file '%s/%s': invalid duration '%s' for '%s' keyword",
1235 line, path, file, value, key);
1238 else if(!strcasecmp(key, "absolute") || !strcasecmp(key, "abs") || !strcasecmp(key, "absolute_sum")) {
1239 *options |= RRDR_OPTION_ABSOLUTE;
1241 else if(!strcasecmp(key, "min2max")) {
1242 *options |= RRDR_OPTION_MIN2MAX;
1244 else if(!strcasecmp(key, "null2zero")) {
1245 *options |= RRDR_OPTION_NULL2ZERO;
1247 else if(!strcasecmp(key, "percentage")) {
1248 *options |= RRDR_OPTION_PERCENTAGE;
1250 else if(!strcasecmp(key, "unaligned")) {
1251 *options |= RRDR_OPTION_NOT_ALIGNED;
1253 else if(!strcasecmp(key, "of")) {
1254 if(*s && strcasecmp(s, "all"))
1255 *dimensions = strdupz(s);
1259 error("Health configuration at line %zu of file '%s/%s': unknown keyword '%s'",
1260 line, path, file, key);
1267 static inline char *health_source_file(size_t line, const char *path, const char *filename) {
1268 char buffer[FILENAME_MAX + 1];
1269 snprintfz(buffer, FILENAME_MAX, "%zu@%s/%s", line, path, filename);
1270 return strdupz(buffer);
1273 static inline void strip_quotes(char *s) {
1275 if(*s == '\'' || *s == '"') *s = ' ';
1280 int health_readfile(const char *path, const char *filename) {
1281 debug(D_HEALTH, "Health configuration reading file '%s/%s'", path, filename);
1283 static uint32_t hash_alarm = 0, hash_template = 0, hash_on = 0, hash_calc = 0, hash_green = 0, hash_red = 0, hash_warn = 0, hash_crit = 0, hash_exec = 0, hash_every = 0, hash_lookup = 0, hash_units = 0, hash_info = 0, hash_recipient = 0, hash_delay = 0;
1284 char buffer[HEALTH_CONF_MAX_LINE + 1];
1286 if(unlikely(!hash_alarm)) {
1287 hash_alarm = simple_uhash(HEALTH_ALARM_KEY);
1288 hash_template = simple_uhash(HEALTH_TEMPLATE_KEY);
1289 hash_on = simple_uhash(HEALTH_ON_KEY);
1290 hash_calc = simple_uhash(HEALTH_CALC_KEY);
1291 hash_lookup = simple_uhash(HEALTH_LOOKUP_KEY);
1292 hash_green = simple_uhash(HEALTH_GREEN_KEY);
1293 hash_red = simple_uhash(HEALTH_RED_KEY);
1294 hash_warn = simple_uhash(HEALTH_WARN_KEY);
1295 hash_crit = simple_uhash(HEALTH_CRIT_KEY);
1296 hash_exec = simple_uhash(HEALTH_EXEC_KEY);
1297 hash_every = simple_uhash(HEALTH_EVERY_KEY);
1298 hash_units = simple_hash(HEALTH_UNITS_KEY);
1299 hash_info = simple_hash(HEALTH_INFO_KEY);
1300 hash_recipient = simple_hash(HEALTH_RECIPIENT_KEY);
1301 hash_delay = simple_uhash(HEALTH_DELAY_KEY);
1304 snprintfz(buffer, HEALTH_CONF_MAX_LINE, "%s/%s", path, filename);
1305 FILE *fp = fopen(buffer, "r");
1307 error("Health configuration cannot read file '%s'.", buffer);
1312 RRDCALCTEMPLATE *rt = NULL;
1314 size_t line = 0, append = 0;
1316 while((s = fgets(&buffer[append], (int)(HEALTH_CONF_MAX_LINE - append), fp)) || append) {
1317 int stop_appending = !s;
1319 // info("Line %zu of file '%s/%s': '%s'", line, path, filename, s);
1322 // info("Trimmed line %zu of file '%s/%s': '%s'", line, path, filename, s);
1325 if(!stop_appending && s[append - 1] == '\\') {
1326 s[append - 1] = ' ';
1327 append = &s[append] - buffer;
1328 if(append < HEALTH_CONF_MAX_LINE)
1331 error("Health configuration has too long muli-line at line %zu of file '%s/%s'.", line, path, filename);
1337 while(*s && *s != ':') s++;
1339 error("Health configuration has invalid line %zu of file '%s/%s'. It does not contain a ':'. Ignoring it.", line, path, filename);
1347 value = trim(value);
1350 error("Health configuration has invalid line %zu of file '%s/%s'. Keyword is empty. Ignoring it.", line, path, filename);
1355 error("Health configuration has invalid line %zu of file '%s/%s'. value is empty. Ignoring it.", line, path, filename);
1359 // info("Health file '%s/%s', key '%s', value '%s'", path, filename, key, value);
1360 uint32_t hash = simple_uhash(key);
1362 if(hash == hash_alarm && !strcasecmp(key, HEALTH_ALARM_KEY)) {
1363 if(rc && !rrdcalc_add_alarm_from_config(&localhost, rc))
1364 rrdcalc_free(&localhost, rc);
1367 if (!rrdcalctemplate_add_template_from_config(&localhost, rt))
1368 rrdcalctemplate_free(&localhost, rt);
1372 rc = callocz(1, sizeof(RRDCALC));
1373 rc->next_event_id = 1;
1374 rc->name = strdupz(value);
1375 rc->hash = simple_hash(rc->name);
1376 rc->source = health_source_file(line, path, filename);
1380 rc->old_value = NAN;
1381 rc->delay_multiplier = 1.0;
1383 if(rrdvar_fix_name(rc->name))
1384 error("Health configuration renamed alarm '%s' to '%s'", value, rc->name);
1386 else if(hash == hash_template && !strcasecmp(key, HEALTH_TEMPLATE_KEY)) {
1388 if(!rrdcalc_add_alarm_from_config(&localhost, rc))
1389 rrdcalc_free(&localhost, rc);
1393 if(rt && !rrdcalctemplate_add_template_from_config(&localhost, rt))
1394 rrdcalctemplate_free(&localhost, rt);
1396 rt = callocz(1, sizeof(RRDCALCTEMPLATE));
1397 rt->name = strdupz(value);
1398 rt->hash_name = simple_hash(rt->name);
1399 rt->source = health_source_file(line, path, filename);
1402 rt->delay_multiplier = 1.0;
1404 if(rrdvar_fix_name(rt->name))
1405 error("Health configuration renamed template '%s' to '%s'", value, rt->name);
1408 if(hash == hash_on && !strcasecmp(key, HEALTH_ON_KEY)) {
1410 if(strcmp(rc->chart, value))
1411 info("Health configuration at line %zu of file '%s/%s' for alarm '%s' has key '%s' twice, once with value '%s' and later with value '%s'. Using ('%s').",
1412 line, path, filename, rc->name, key, rc->chart, value, value);
1416 rc->chart = strdupz(value);
1417 rc->hash_chart = simple_hash(rc->chart);
1419 else if(hash == hash_lookup && !strcasecmp(key, HEALTH_LOOKUP_KEY)) {
1420 health_parse_db_lookup(line, path, filename, value, &rc->group, &rc->after, &rc->before,
1422 &rc->options, &rc->dimensions);
1424 else if(hash == hash_every && !strcasecmp(key, HEALTH_EVERY_KEY)) {
1425 if(!health_parse_duration(value, &rc->update_every))
1426 info("Health configuration at line %zu of file '%s/%s' for alarm '%s' at key '%s' cannot parse duration: '%s'.",
1427 line, path, filename, rc->name, key, value);
1429 else if(hash == hash_green && !strcasecmp(key, HEALTH_GREEN_KEY)) {
1431 rc->green = strtold(value, &e);
1433 info("Health configuration at line %zu of file '%s/%s' for alarm '%s' at key '%s' leaves this string unmatched: '%s'.",
1434 line, path, filename, rc->name, key, e);
1437 else if(hash == hash_red && !strcasecmp(key, HEALTH_RED_KEY)) {
1439 rc->red = strtold(value, &e);
1441 info("Health configuration at line %zu of file '%s/%s' for alarm '%s' at key '%s' leaves this string unmatched: '%s'.",
1442 line, path, filename, rc->name, key, e);
1445 else if(hash == hash_calc && !strcasecmp(key, HEALTH_CALC_KEY)) {
1446 const char *failed_at = NULL;
1448 rc->calculation = expression_parse(value, &failed_at, &error);
1449 if(!rc->calculation) {
1450 error("Health configuration at line %zu of file '%s/%s' for alarm '%s' at key '%s' has unparse-able expression '%s': %s at '%s'",
1451 line, path, filename, rc->name, key, value, expression_strerror(error), failed_at);
1454 else if(hash == hash_warn && !strcasecmp(key, HEALTH_WARN_KEY)) {
1455 const char *failed_at = NULL;
1457 rc->warning = expression_parse(value, &failed_at, &error);
1459 error("Health configuration at line %zu of file '%s/%s' for alarm '%s' at key '%s' has unparse-able expression '%s': %s at '%s'",
1460 line, path, filename, rc->name, key, value, expression_strerror(error), failed_at);
1463 else if(hash == hash_crit && !strcasecmp(key, HEALTH_CRIT_KEY)) {
1464 const char *failed_at = NULL;
1466 rc->critical = expression_parse(value, &failed_at, &error);
1468 error("Health configuration at line %zu of file '%s/%s' for alarm '%s' at key '%s' has unparse-able expression '%s': %s at '%s'",
1469 line, path, filename, rc->name, key, value, expression_strerror(error), failed_at);
1472 else if(hash == hash_exec && !strcasecmp(key, HEALTH_EXEC_KEY)) {
1474 if(strcmp(rc->exec, value))
1475 info("Health configuration at line %zu of file '%s/%s' for alarm '%s' has key '%s' twice, once with value '%s' and later with value '%s'. Using ('%s').",
1476 line, path, filename, rc->name, key, rc->exec, value, value);
1480 rc->exec = strdupz(value);
1482 else if(hash == hash_recipient && !strcasecmp(key, HEALTH_RECIPIENT_KEY)) {
1484 if(strcmp(rc->recipient, value))
1485 info("Health configuration at line %zu of file '%s/%s' for alarm '%s' has key '%s' twice, once with value '%s' and later with value '%s'. Using ('%s').",
1486 line, path, filename, rc->name, key, rc->recipient, value, value);
1488 freez(rc->recipient);
1490 rc->recipient = strdupz(value);
1492 else if(hash == hash_units && !strcasecmp(key, HEALTH_UNITS_KEY)) {
1494 if(strcmp(rc->units, value))
1495 info("Health configuration at line %zu of file '%s/%s' for alarm '%s' has key '%s' twice, once with value '%s' and later with value '%s'. Using ('%s').",
1496 line, path, filename, rc->name, key, rc->units, value, value);
1500 rc->units = strdupz(value);
1501 strip_quotes(rc->units);
1503 else if(hash == hash_info && !strcasecmp(key, HEALTH_INFO_KEY)) {
1505 if(strcmp(rc->info, value))
1506 info("Health configuration at line %zu of file '%s/%s' for alarm '%s' has key '%s' twice, once with value '%s' and later with value '%s'. Using ('%s').",
1507 line, path, filename, rc->name, key, rc->info, value, value);
1511 rc->info = strdupz(value);
1512 strip_quotes(rc->info);
1514 else if(hash == hash_delay && !strcasecmp(key, HEALTH_DELAY_KEY)) {
1515 health_parse_delay(line, path, filename, value, &rc->delay_up_duration, &rc->delay_down_duration, &rc->delay_max_duration, &rc->delay_multiplier);
1518 error("Health configuration at line %zu of file '%s/%s' for alarm '%s' has unknown key '%s'.",
1519 line, path, filename, rc->name, key);
1523 if(hash == hash_on && !strcasecmp(key, HEALTH_ON_KEY)) {
1525 if(strcmp(rt->context, value))
1526 info("Health configuration at line %zu of file '%s/%s' for template '%s' has key '%s' twice, once with value '%s' and later with value '%s'. Using ('%s').",
1527 line, path, filename, rt->name, key, rt->context, value, value);
1531 rt->context = strdupz(value);
1532 rt->hash_context = simple_hash(rt->context);
1534 else if(hash == hash_lookup && !strcasecmp(key, HEALTH_LOOKUP_KEY)) {
1535 health_parse_db_lookup(line, path, filename, value, &rt->group, &rt->after, &rt->before,
1537 &rt->options, &rt->dimensions);
1539 else if(hash == hash_every && !strcasecmp(key, HEALTH_EVERY_KEY)) {
1540 if(!health_parse_duration(value, &rt->update_every))
1541 info("Health configuration at line %zu of file '%s/%s' for template '%s' at key '%s' cannot parse duration: '%s'.",
1542 line, path, filename, rt->name, key, value);
1544 else if(hash == hash_green && !strcasecmp(key, HEALTH_GREEN_KEY)) {
1546 rt->green = strtold(value, &e);
1548 info("Health configuration at line %zu of file '%s/%s' for template '%s' at key '%s' leaves this string unmatched: '%s'.",
1549 line, path, filename, rt->name, key, e);
1552 else if(hash == hash_red && !strcasecmp(key, HEALTH_RED_KEY)) {
1554 rt->red = strtold(value, &e);
1556 info("Health configuration at line %zu of file '%s/%s' for template '%s' at key '%s' leaves this string unmatched: '%s'.",
1557 line, path, filename, rt->name, key, e);
1560 else if(hash == hash_calc && !strcasecmp(key, HEALTH_CALC_KEY)) {
1561 const char *failed_at = NULL;
1563 rt->calculation = expression_parse(value, &failed_at, &error);
1564 if(!rt->calculation) {
1565 error("Health configuration at line %zu of file '%s/%s' for template '%s' at key '%s' has unparse-able expression '%s': %s at '%s'",
1566 line, path, filename, rt->name, key, value, expression_strerror(error), failed_at);
1569 else if(hash == hash_warn && !strcasecmp(key, HEALTH_WARN_KEY)) {
1570 const char *failed_at = NULL;
1572 rt->warning = expression_parse(value, &failed_at, &error);
1574 error("Health configuration at line %zu of file '%s/%s' for template '%s' at key '%s' has unparse-able expression '%s': %s at '%s'",
1575 line, path, filename, rt->name, key, value, expression_strerror(error), failed_at);
1578 else if(hash == hash_crit && !strcasecmp(key, HEALTH_CRIT_KEY)) {
1579 const char *failed_at = NULL;
1581 rt->critical = expression_parse(value, &failed_at, &error);
1583 error("Health configuration at line %zu of file '%s/%s' for template '%s' at key '%s' has unparse-able expression '%s': %s at '%s'",
1584 line, path, filename, rt->name, key, value, expression_strerror(error), failed_at);
1587 else if(hash == hash_exec && !strcasecmp(key, HEALTH_EXEC_KEY)) {
1589 if(strcmp(rt->exec, value))
1590 info("Health configuration at line %zu of file '%s/%s' for template '%s' has key '%s' twice, once with value '%s' and later with value '%s'. Using ('%s').",
1591 line, path, filename, rt->name, key, rt->exec, value, value);
1595 rt->exec = strdupz(value);
1597 else if(hash == hash_recipient && !strcasecmp(key, HEALTH_RECIPIENT_KEY)) {
1599 if(strcmp(rt->recipient, value))
1600 info("Health configuration at line %zu of file '%s/%s' for template '%s' has key '%s' twice, once with value '%s' and later with value '%s'. Using ('%s').",
1601 line, path, filename, rt->name, key, rt->recipient, value, value);
1603 freez(rt->recipient);
1605 rt->recipient = strdupz(value);
1607 else if(hash == hash_units && !strcasecmp(key, HEALTH_UNITS_KEY)) {
1609 if(strcmp(rt->units, value))
1610 info("Health configuration at line %zu of file '%s/%s' for template '%s' has key '%s' twice, once with value '%s' and later with value '%s'. Using ('%s').",
1611 line, path, filename, rt->name, key, rt->units, value, value);
1615 rt->units = strdupz(value);
1616 strip_quotes(rt->units);
1618 else if(hash == hash_info && !strcasecmp(key, HEALTH_INFO_KEY)) {
1620 if(strcmp(rt->info, value))
1621 info("Health configuration at line %zu of file '%s/%s' for template '%s' has key '%s' twice, once with value '%s' and later with value '%s'. Using ('%s').",
1622 line, path, filename, rt->name, key, rt->info, value, value);
1626 rt->info = strdupz(value);
1627 strip_quotes(rt->info);
1629 else if(hash == hash_delay && !strcasecmp(key, HEALTH_DELAY_KEY)) {
1630 health_parse_delay(line, path, filename, value, &rt->delay_up_duration, &rt->delay_down_duration, &rt->delay_max_duration, &rt->delay_multiplier);
1633 error("Health configuration at line %zu of file '%s/%s' for template '%s' has unknown key '%s'.",
1634 line, path, filename, rt->name, key);
1638 error("Health configuration at line %zu of file '%s/%s' has unknown key '%s'. Expected either '" HEALTH_ALARM_KEY "' or '" HEALTH_TEMPLATE_KEY "'.",
1639 line, path, filename, key);
1643 if(rc && !rrdcalc_add_alarm_from_config(&localhost, rc))
1644 rrdcalc_free(&localhost, rc);
1646 if(rt && !rrdcalctemplate_add_template_from_config(&localhost, rt))
1647 rrdcalctemplate_free(&localhost, rt);
1653 void health_readdir(const char *path) {
1654 size_t pathlen = strlen(path);
1656 debug(D_HEALTH, "Health configuration reading directory '%s'", path);
1658 DIR *dir = opendir(path);
1660 error("Health configuration cannot open directory '%s'.", path);
1664 struct dirent *de = NULL;
1665 while ((de = readdir(dir))) {
1666 size_t len = strlen(de->d_name);
1668 if(de->d_type == DT_DIR
1670 (de->d_name[0] == '.' && de->d_name[1] == '\0')
1671 || (de->d_name[0] == '.' && de->d_name[1] == '.' && de->d_name[2] == '\0')
1673 debug(D_HEALTH, "Ignoring directory '%s'", de->d_name);
1677 else if(de->d_type == DT_DIR) {
1678 char *s = mallocz(pathlen + strlen(de->d_name) + 2);
1681 strcat(s, de->d_name);
1687 else if((de->d_type == DT_LNK || de->d_type == DT_REG || de->d_type == DT_UNKNOWN) &&
1688 len > 5 && !strcmp(&de->d_name[len - 5], ".conf")) {
1689 health_readfile(path, de->d_name);
1692 else debug(D_HEALTH, "Ignoring file '%s'", de->d_name);
1698 static inline char *health_config_dir(void) {
1699 char buffer[FILENAME_MAX + 1];
1700 snprintfz(buffer, FILENAME_MAX, "%s/health.d", config_get("global", "config directory", CONFIG_DIR));
1701 return config_get("health", "health configuration directory", buffer);
1704 void health_init(void) {
1705 debug(D_HEALTH, "Health configuration initializing");
1707 if(!(health_enabled = config_get_boolean("health", "enabled", 1))) {
1708 debug(D_HEALTH, "Health is disabled.");
1712 char *path = health_config_dir();
1715 char buffer[FILENAME_MAX + 1];
1716 snprintfz(buffer, FILENAME_MAX, "%s/alarm-notify.sh", config_get("global", "plugins directory", PLUGINS_DIR));
1717 health_default_exec = config_get("health", "script to execute on alarm", buffer);
1720 long n = config_get_number("health", "in memory max health log entries", (long)localhost.health_log.max);
1722 error("Health configuration has invalid max log entries %ld. Using default %u", n, localhost.health_log.max);
1723 config_set_number("health", "in memory max health log entries", (long)localhost.health_log.max);
1725 else localhost.health_log.max = (unsigned int)n;
1727 rrdhost_rwlock(&localhost);
1728 health_readdir(path);
1729 rrdhost_unlock(&localhost);
1732 // ----------------------------------------------------------------------------
1735 static inline void health_string2json(BUFFER *wb, const char *prefix, const char *label, const char *value, const char *suffix) {
1737 buffer_sprintf(wb, "%s\"%s\":\"%s\"%s", prefix, label, value, suffix);
1739 buffer_sprintf(wb, "%s\"%s\":null%s", prefix, label, suffix);
1742 static inline void health_alarm_entry2json_nolock(BUFFER *wb, ALARM_ENTRY *ae, RRDHOST *host) {
1743 buffer_sprintf(wb, "\n\t{\n"
1744 "\t\t\"hostname\": \"%s\",\n"
1745 "\t\t\"unique_id\": %u,\n"
1746 "\t\t\"alarm_id\": %u,\n"
1747 "\t\t\"alarm_event_id\": %u,\n"
1748 "\t\t\"name\": \"%s\",\n"
1749 "\t\t\"chart\": \"%s\",\n"
1750 "\t\t\"family\": \"%s\",\n"
1751 "\t\t\"processed\": %s,\n"
1752 "\t\t\"updated\": %s,\n"
1753 "\t\t\"exec_run\": %lu,\n"
1754 "\t\t\"exec_failed\": %s,\n"
1755 "\t\t\"exec\": \"%s\",\n"
1756 "\t\t\"recipient\": \"%s\",\n"
1757 "\t\t\"exec_code\": %d,\n"
1758 "\t\t\"source\": \"%s\",\n"
1759 "\t\t\"units\": \"%s\",\n"
1760 "\t\t\"info\": \"%s\",\n"
1761 "\t\t\"when\": %lu,\n"
1762 "\t\t\"duration\": %lu,\n"
1763 "\t\t\"non_clear_duration\": %lu,\n"
1764 "\t\t\"status\": \"%s\",\n"
1765 "\t\t\"old_status\": \"%s\",\n"
1766 "\t\t\"delay\": %d,\n"
1767 "\t\t\"delay_up_to_timestamp\": %lu,\n",
1775 (ae->notifications & HEALTH_ENTRY_NOTIFICATIONS_PROCESSED)?"true":"false",
1776 (ae->notifications & HEALTH_ENTRY_NOTIFICATIONS_UPDATED)?"true":"false",
1777 (unsigned long)ae->exec_run_timestamp,
1778 (ae->notifications & HEALTH_ENTRY_NOTIFICATIONS_EXEC_FAILED)?"true":"false",
1779 ae->exec?ae->exec:health_default_exec,
1780 ae->recipient?ae->recipient:health_default_recipient,
1783 ae->units?ae->units:"",
1784 ae->info?ae->info:"",
1785 (unsigned long)ae->when,
1786 (unsigned long)ae->duration,
1787 (unsigned long)ae->non_clear_duration,
1788 rrdcalc_status2string(ae->new_status),
1789 rrdcalc_status2string(ae->old_status),
1791 (unsigned long)ae->delay_up_to_timestamp
1794 buffer_strcat(wb, "\t\t\"value\":");
1795 buffer_rrd_value(wb, ae->new_value);
1796 buffer_strcat(wb, ",\n");
1798 buffer_strcat(wb, "\t\t\"old_value\":");
1799 buffer_rrd_value(wb, ae->old_value);
1800 buffer_strcat(wb, "\n");
1802 buffer_strcat(wb, "\t}");
1805 void health_alarm_log2json(RRDHOST *host, BUFFER *wb, uint32_t after) {
1806 pthread_rwlock_rdlock(&host->health_log.alarm_log_rwlock);
1808 buffer_strcat(wb, "[");
1810 unsigned int max = host->health_log.max;
1811 unsigned int count = 0;
1813 for(ae = host->health_log.alarms; ae && count < max ; count++, ae = ae->next) {
1814 if(ae->unique_id > after) {
1815 if(likely(count)) buffer_strcat(wb, ",");
1816 health_alarm_entry2json_nolock(wb, ae, host);
1820 buffer_strcat(wb, "\n]\n");
1822 pthread_rwlock_unlock(&host->health_log.alarm_log_rwlock);
1825 static inline void health_rrdcalc2json_nolock(BUFFER *wb, RRDCALC *rc) {
1827 "\t\t\"%s.%s\": {\n"
1828 "\t\t\t\"id\": %lu,\n"
1829 "\t\t\t\"name\": \"%s\",\n"
1830 "\t\t\t\"chart\": \"%s\",\n"
1831 "\t\t\t\"family\": \"%s\",\n"
1832 "\t\t\t\"active\": %s,\n"
1833 "\t\t\t\"exec\": \"%s\",\n"
1834 "\t\t\t\"recipient\": \"%s\",\n"
1835 "\t\t\t\"source\": \"%s\",\n"
1836 "\t\t\t\"units\": \"%s\",\n"
1837 "\t\t\t\"info\": \"%s\",\n"
1838 "\t\t\t\"status\": \"%s\",\n"
1839 "\t\t\t\"last_status_change\": %lu,\n"
1840 "\t\t\t\"last_updated\": %lu,\n"
1841 "\t\t\t\"next_update\": %lu,\n"
1842 "\t\t\t\"update_every\": %d,\n"
1843 "\t\t\t\"delay_up_duration\": %d,\n"
1844 "\t\t\t\"delay_down_duration\": %d,\n"
1845 "\t\t\t\"delay_max_duration\": %d,\n"
1846 "\t\t\t\"delay_multiplier\": %f,\n"
1847 "\t\t\t\"delay\": %d,\n"
1848 "\t\t\t\"delay_up_to_timestamp\": %lu,\n"
1849 , rc->chart, rc->name
1850 , (unsigned long)rc->id
1853 , (rc->rrdset && rc->rrdset->family)?rc->rrdset->family:""
1854 , (rc->rrdset)?"true":"false"
1855 , rc->exec?rc->exec:health_default_exec
1856 , rc->recipient?rc->recipient:health_default_recipient
1858 , rc->units?rc->units:""
1859 , rc->info?rc->info:""
1860 , rrdcalc_status2string(rc->status)
1861 , (unsigned long)rc->last_status_change
1862 , (unsigned long)rc->last_updated
1863 , (unsigned long)rc->next_update
1865 , rc->delay_up_duration
1866 , rc->delay_down_duration
1867 , rc->delay_max_duration
1868 , rc->delay_multiplier
1870 , (unsigned long)rc->delay_up_to_timestamp
1873 if(RRDCALC_HAS_DB_LOOKUP(rc)) {
1874 if(rc->dimensions && *rc->dimensions)
1875 health_string2json(wb, "\t\t\t", "lookup_dimensions", rc->dimensions, ",\n");
1878 "\t\t\t\"db_after\": %lu,\n"
1879 "\t\t\t\"db_before\": %lu,\n"
1880 "\t\t\t\"lookup_method\": \"%s\",\n"
1881 "\t\t\t\"lookup_after\": %d,\n"
1882 "\t\t\t\"lookup_before\": %d,\n"
1883 "\t\t\t\"lookup_options\": \"",
1884 (unsigned long) rc->db_after,
1885 (unsigned long) rc->db_before,
1886 group_method2string(rc->group),
1890 buffer_data_options2string(wb, rc->options);
1891 buffer_strcat(wb, "\",\n");
1894 if(rc->calculation) {
1895 health_string2json(wb, "\t\t\t", "calc", rc->calculation->source, ",\n");
1896 health_string2json(wb, "\t\t\t", "calc_parsed", rc->calculation->parsed_as, ",\n");
1900 health_string2json(wb, "\t\t\t", "warn", rc->warning->source, ",\n");
1901 health_string2json(wb, "\t\t\t", "warn_parsed", rc->warning->parsed_as, ",\n");
1905 health_string2json(wb, "\t\t\t", "crit", rc->critical->source, ",\n");
1906 health_string2json(wb, "\t\t\t", "crit_parsed", rc->critical->parsed_as, ",\n");
1909 buffer_strcat(wb, "\t\t\t\"green\":");
1910 buffer_rrd_value(wb, rc->green);
1911 buffer_strcat(wb, ",\n");
1913 buffer_strcat(wb, "\t\t\t\"red\":");
1914 buffer_rrd_value(wb, rc->red);
1915 buffer_strcat(wb, ",\n");
1917 buffer_strcat(wb, "\t\t\t\"value\":");
1918 buffer_rrd_value(wb, rc->value);
1919 buffer_strcat(wb, "\n");
1921 buffer_strcat(wb, "\t\t}");
1924 //void health_rrdcalctemplate2json_nolock(BUFFER *wb, RRDCALCTEMPLATE *rt) {
1928 void health_alarms2json(RRDHOST *host, BUFFER *wb, int all) {
1931 rrdhost_rdlock(&localhost);
1932 buffer_sprintf(wb, "{\n\t\"hostname\": \"%s\","
1933 "\n\t\"latest_alarm_log_unique_id\": %u,"
1934 "\n\t\"status\": %s,"
1936 "\n\t\"alarms\": {\n",
1938 (host->health_log.next_log_id > 0)?(host->health_log.next_log_id - 1):0,
1939 health_enabled?"true":"false",
1940 (unsigned long)time(NULL));
1943 for(i = 0, rc = host->alarms; rc ; rc = rc->next) {
1944 if(unlikely(!rc->rrdset || !rc->rrdset->last_collected_time.tv_sec))
1947 if(likely(!all && !(rc->status == RRDCALC_STATUS_WARNING || rc->status == RRDCALC_STATUS_CRITICAL)))
1950 if(likely(i)) buffer_strcat(wb, ",\n");
1951 health_rrdcalc2json_nolock(wb, rc);
1955 // buffer_strcat(wb, "\n\t},\n\t\"templates\": {");
1956 // RRDCALCTEMPLATE *rt;
1957 // for(rt = host->templates; rt ; rt = rt->next)
1958 // health_rrdcalctemplate2json_nolock(wb, rt);
1960 buffer_strcat(wb, "\n\t}\n}\n");
1961 rrdhost_unlock(&localhost);
1965 // ----------------------------------------------------------------------------
1966 // re-load health configuration
1968 static inline void health_free_all_nolock(RRDHOST *host) {
1969 while(host->templates)
1970 rrdcalctemplate_free(host, host->templates);
1973 rrdcalc_free(host, host->alarms);
1976 void health_reload(void) {
1977 if(!health_enabled) {
1978 error("Health reload is requested, but health is not enabled.");
1982 char *path = health_config_dir();
1984 // free all running alarms
1985 rrdhost_rwlock(&localhost);
1986 health_free_all_nolock(&localhost);
1987 rrdhost_unlock(&localhost);
1989 // invalidate all previous entries in the alarm log
1991 for(t = localhost.health_log.alarms ; t ; t = t->next) {
1992 if(t->new_status != RRDCALC_STATUS_REMOVED)
1993 t->notifications |= HEALTH_ENTRY_NOTIFICATIONS_UPDATED;
1996 // reset all thresholds to all charts
1998 for(st = localhost.rrdset_root; st ; st = st->next) {
2003 // load the new alarms
2004 rrdhost_rwlock(&localhost);
2005 health_readdir(path);
2006 rrdhost_unlock(&localhost);
2008 // link the loaded alarms to their charts
2009 for(st = localhost.rrdset_root; st ; st = st->next) {
2010 rrdhost_rwlock(&localhost);
2012 rrdsetcalc_link_matching(st);
2013 rrdcalctemplate_link_matching(st);
2015 rrdhost_unlock(&localhost);
2020 // ----------------------------------------------------------------------------
2021 // health main thread and friends
2023 static inline int rrdcalc_value2status(calculated_number n) {
2024 if(isnan(n)) return RRDCALC_STATUS_UNDEFINED;
2025 if(n) return RRDCALC_STATUS_RAISED;
2026 return RRDCALC_STATUS_CLEAR;
2029 static inline void health_alarm_execute(RRDHOST *host, ALARM_ENTRY *ae) {
2030 ae->notifications |= HEALTH_ENTRY_NOTIFICATIONS_PROCESSED;
2032 // find the previous notification for the same alarm
2034 for(t = ae->next; t ;t = t->next) {
2035 if(t->alarm_id == ae->alarm_id && t->notifications & HEALTH_ENTRY_NOTIFICATIONS_EXEC_RUN)
2039 if(t && t->new_status == ae->new_status) {
2040 // don't send the same notification again
2041 info("Health not sending again notification for alarm '%s.%s' status %s", ae->chart, ae->name, rrdcalc_status2string(ae->new_status));
2045 if(ae->old_status == RRDCALC_STATUS_UNDEFINED && ae->new_status == RRDCALC_STATUS_UNINITIALIZED
2046 || ae->old_status == RRDCALC_STATUS_UNINITIALIZED && ae->new_status == RRDCALC_STATUS_CLEAR) {
2047 info("Health not sending notification for first initialization of alarm '%s.%s' status %s", ae->chart, ae->name, rrdcalc_status2string(ae->new_status));
2051 char buffer[FILENAME_MAX + 1];
2054 const char *exec = ae->exec;
2055 if(!exec) exec = health_default_exec;
2057 const char *recipient = ae->recipient;
2058 if(!recipient) recipient = health_default_recipient;
2060 snprintfz(buffer, FILENAME_MAX, "exec %s '%s' '%s' '%u' '%u' '%u' '%lu' '%s' '%s' '%s' '%s' '%s' '%0.0Lf' '%0.0Lf' '%s' '%u' '%u' '%s' '%s'",
2067 (unsigned long)ae->when,
2069 ae->chart?ae->chart:"NOCAHRT",
2070 ae->family?ae->family:"NOFAMILY",
2071 rrdcalc_status2string(ae->new_status),
2072 rrdcalc_status2string(ae->old_status),
2075 ae->source?ae->source:"UNKNOWN",
2076 (uint32_t)ae->duration,
2077 (uint32_t)ae->non_clear_duration,
2078 ae->units?ae->units:"",
2079 ae->info?ae->info:""
2082 ae->notifications |= HEALTH_ENTRY_NOTIFICATIONS_EXEC_RUN;
2083 ae->exec_run_timestamp = time(NULL);
2085 debug(D_HEALTH, "executing command '%s'", buffer);
2086 FILE *fp = mypopen(buffer, &command_pid);
2088 error("HEALTH: Cannot popen(\"%s\", \"r\").", buffer);
2091 debug(D_HEALTH, "HEALTH reading from command");
2092 char *s = fgets(buffer, FILENAME_MAX, fp);
2094 ae->exec_code = mypclose(fp, command_pid);
2095 debug(D_HEALTH, "done executing command - returned with code %d", ae->exec_code);
2097 if(ae->exec_code != 0)
2098 ae->notifications |= HEALTH_ENTRY_NOTIFICATIONS_EXEC_FAILED;
2101 static inline void health_process_notifications(RRDHOST *host, ALARM_ENTRY *ae) {
2102 info("Health alarm '%s.%s' = %0.2Lf - changed status from %s to %s",
2103 ae->chart?ae->chart:"NOCHART", ae->name,
2105 rrdcalc_status2string(ae->old_status),
2106 rrdcalc_status2string(ae->new_status)
2109 health_alarm_execute(host, ae);
2112 static inline void health_alarm_log_process(RRDHOST *host) {
2113 static uint32_t stop_at_id = 0;
2114 uint32_t first_waiting = (host->health_log.alarms)?host->health_log.alarms->unique_id:0;
2115 time_t now = time(NULL);
2117 pthread_rwlock_rdlock(&host->health_log.alarm_log_rwlock);
2120 for(ae = host->health_log.alarms; ae && ae->unique_id >= stop_at_id ; ae = ae->next) {
2122 !(ae->notifications & HEALTH_ENTRY_NOTIFICATIONS_PROCESSED) &&
2123 !(ae->notifications & HEALTH_ENTRY_NOTIFICATIONS_UPDATED)
2126 if(unlikely(ae->unique_id < first_waiting))
2127 first_waiting = ae->unique_id;
2129 if(likely(now > ae->delay_up_to_timestamp))
2130 health_process_notifications(host, ae);
2134 // remember this for the next iteration
2135 stop_at_id = first_waiting;
2137 pthread_rwlock_unlock(&host->health_log.alarm_log_rwlock);
2139 if(host->health_log.count <= host->health_log.max)
2142 // cleanup excess entries in the log
2143 pthread_rwlock_wrlock(&host->health_log.alarm_log_rwlock);
2145 ALARM_ENTRY *last = NULL;
2146 unsigned int count = host->health_log.max * 2 / 3;
2147 for(ae = host->health_log.alarms; ae && count ; count--, last = ae, ae = ae->next) ;
2149 if(ae && last && last->next == ae)
2155 debug(D_HEALTH, "Health removing alarm log entry with id: %u", ae->unique_id);
2157 ALARM_ENTRY *t = ae->next;
2163 freez(ae->recipient);
2170 host->health_log.count--;
2173 pthread_rwlock_unlock(&host->health_log.alarm_log_rwlock);
2176 static inline int rrdcalc_isrunnable(RRDCALC *rc, time_t now, time_t *next_run) {
2177 if (unlikely(!rc->rrdset)) {
2178 debug(D_HEALTH, "Health not running alarm '%s.%s'. It is not linked to a chart.", rc->chart?rc->chart:"NOCHART", rc->name);
2182 if (unlikely(!rc->rrdset->last_collected_time.tv_sec)) {
2183 debug(D_HEALTH, "Health not running alarm '%s.%s'. Chart is not yet collected.", rc->chart?rc->chart:"NOCHART", rc->name);
2187 if (unlikely(!rc->update_every)) {
2188 debug(D_HEALTH, "Health not running alarm '%s.%s'. It does not have an update frequency", rc->chart?rc->chart:"NOCHART", rc->name);
2192 if (unlikely(rc->next_update > now)) {
2193 if (unlikely(*next_run > rc->next_update))
2194 *next_run = rc->next_update;
2196 debug(D_HEALTH, "Health not examining alarm '%s.%s' yet (will do in %d secs).", rc->chart?rc->chart:"NOCHART", rc->name, (int) (rc->next_update - now));
2201 // we should check that the DB lookup is possible
2203 // - the duration of the chart includes the required timeframe
2204 // we SHOULD NOT check the dimensions - there might be alarms that refer non-existing dimensions (e.g. cpu steal)
2209 void *health_main(void *ptr) {
2212 info("HEALTH thread created with task id %d", gettid());
2214 if(pthread_setcanceltype(PTHREAD_CANCEL_DEFERRED, NULL) != 0)
2215 error("Cannot set pthread cancel type to DEFERRED.");
2217 if(pthread_setcancelstate(PTHREAD_CANCEL_ENABLE, NULL) != 0)
2218 error("Cannot set pthread cancel state to ENABLE.");
2220 int min_run_every = (int)config_get_number("health", "run at least every seconds", 10);
2221 if(min_run_every < 1) min_run_every = 1;
2223 BUFFER *wb = buffer_create(100);
2225 unsigned int loop = 0;
2226 while(health_enabled) {
2228 debug(D_HEALTH, "Health monitoring iteration no %u started", loop);
2230 int oldstate, runnable = 0;
2231 time_t now = time(NULL);
2232 time_t next_run = now + min_run_every;
2235 if (unlikely(pthread_setcancelstate(PTHREAD_CANCEL_DISABLE, &oldstate) != 0))
2236 error("Cannot set pthread cancel state to DISABLE.");
2238 rrdhost_rdlock(&localhost);
2240 // the first loop is to lookup values from the db
2241 for (rc = localhost.alarms; rc; rc = rc->next) {
2242 if (unlikely(!rrdcalc_isrunnable(rc, now, &next_run)))
2246 rc->old_value = rc->value;
2248 // 1. if there is database lookup, do it
2249 // 2. if there is calculation expression, run it
2251 if (unlikely(RRDCALC_HAS_DB_LOOKUP(rc))) {
2252 time_t old_db_timestamp = rc->db_before;
2253 int value_is_null = 0;
2255 int ret = rrd2value(rc->rrdset, wb, &rc->value,
2256 rc->dimensions, 1, rc->after, rc->before, rc->group,
2257 rc->options, &rc->db_after, &rc->db_before, &value_is_null);
2259 if (unlikely(ret != 200)) {
2260 // database lookup failed
2263 debug(D_HEALTH, "Health alarm '%s.%s': database lookup returned error %d", rc->chart?rc->chart:"NOCHART", rc->name, ret);
2265 if (unlikely(!(rc->rrdcalc_flags & RRDCALC_FLAG_DB_ERROR))) {
2266 rc->rrdcalc_flags |= RRDCALC_FLAG_DB_ERROR;
2267 error("Health alarm '%s.%s': database lookup returned error %d", rc->chart?rc->chart:"NOCHART", rc->name, ret);
2270 else if (unlikely(rc->rrdcalc_flags & RRDCALC_FLAG_DB_ERROR))
2271 rc->rrdcalc_flags &= ~RRDCALC_FLAG_DB_ERROR;
2273 if (unlikely(old_db_timestamp == rc->db_before)) {
2274 // database is stale
2276 debug(D_HEALTH, "Health alarm '%s.%s': database is stale", rc->chart?rc->chart:"NOCHART", rc->name);
2278 if (unlikely(!(rc->rrdcalc_flags & RRDCALC_FLAG_DB_STALE))) {
2279 rc->rrdcalc_flags |= RRDCALC_FLAG_DB_STALE;
2280 error("Health alarm '%s.%s': database is stale", rc->chart?rc->chart:"NOCHART", rc->name);
2283 else if (unlikely(rc->rrdcalc_flags & RRDCALC_FLAG_DB_STALE))
2284 rc->rrdcalc_flags &= ~RRDCALC_FLAG_DB_STALE;
2286 if (unlikely(value_is_null)) {
2287 // collected value is null
2291 debug(D_HEALTH, "Health alarm '%s.%s': database lookup returned empty value (possibly value is not collected yet)",
2292 rc->chart?rc->chart:"NOCHART", rc->name);
2294 if (unlikely(!(rc->rrdcalc_flags & RRDCALC_FLAG_DB_NAN))) {
2295 rc->rrdcalc_flags |= RRDCALC_FLAG_DB_NAN;
2296 error("Health alarm '%s.%s': database lookup returned empty value (possibly value is not collected yet)",
2297 rc->chart?rc->chart:"NOCHART", rc->name);
2300 else if (unlikely(rc->rrdcalc_flags & RRDCALC_FLAG_DB_NAN))
2301 rc->rrdcalc_flags &= ~RRDCALC_FLAG_DB_NAN;
2303 debug(D_HEALTH, "Health alarm '%s.%s': database lookup gave value "
2304 CALCULATED_NUMBER_FORMAT, rc->chart?rc->chart:"NOCHART", rc->name, rc->value);
2307 if(unlikely(rc->calculation)) {
2308 if (unlikely(!expression_evaluate(rc->calculation))) {
2309 // calculation failed
2313 debug(D_HEALTH, "Health alarm '%s.%s': failed to evaluate calculation with error: %s",
2314 rc->chart?rc->chart:"NOCHART", rc->name, buffer_tostring(rc->calculation->error_msg));
2316 if (unlikely(!(rc->rrdcalc_flags & RRDCALC_FLAG_CALC_ERROR))) {
2317 rc->rrdcalc_flags |= RRDCALC_FLAG_CALC_ERROR;
2318 error("Health alarm '%s.%s': failed to evaluate calculation with error: %s",
2319 rc->chart?rc->chart:"NOCHART", rc->name, buffer_tostring(rc->calculation->error_msg));
2323 if (unlikely(rc->rrdcalc_flags & RRDCALC_FLAG_CALC_ERROR))
2324 rc->rrdcalc_flags &= ~RRDCALC_FLAG_CALC_ERROR;
2326 debug(D_HEALTH, "Health alarm '%s.%s': calculation expression gave value "
2327 CALCULATED_NUMBER_FORMAT
2328 ": %s (source: %s)",
2329 rc->chart?rc->chart:"NOCHART", rc->name,
2330 rc->calculation->result,
2331 buffer_tostring(rc->calculation->error_msg),
2335 rc->value = rc->calculation->result;
2339 rrdhost_unlock(&localhost);
2342 rrdhost_rdlock(&localhost);
2344 for (rc = localhost.alarms; rc; rc = rc->next) {
2345 if (unlikely(!rrdcalc_isrunnable(rc, now, &next_run)))
2348 int warning_status = RRDCALC_STATUS_UNDEFINED;
2349 int critical_status = RRDCALC_STATUS_UNDEFINED;
2351 if(likely(rc->warning)) {
2352 if(unlikely(!expression_evaluate(rc->warning))) {
2353 // calculation failed
2355 debug(D_HEALTH, "Health alarm '%s.%s': warning expression failed with error: %s",
2356 rc->chart?rc->chart:"NOCHART", rc->name, buffer_tostring(rc->warning->error_msg));
2358 if (unlikely(!(rc->rrdcalc_flags & RRDCALC_FLAG_WARN_ERROR))) {
2359 rc->rrdcalc_flags |= RRDCALC_FLAG_WARN_ERROR;
2360 error("Health alarm '%s.%s': warning expression failed with error: %s",
2361 rc->chart?rc->chart:"NOCHART", rc->name, buffer_tostring(rc->warning->error_msg));
2365 if(unlikely(rc->rrdcalc_flags & RRDCALC_FLAG_WARN_ERROR))
2366 rc->rrdcalc_flags &= ~RRDCALC_FLAG_WARN_ERROR;
2368 debug(D_HEALTH, "Health alarm '%s.%s': warning expression gave value "
2369 CALCULATED_NUMBER_FORMAT
2370 ": %s (source: %s)",
2371 rc->chart?rc->chart:"NOCHART", rc->name,
2372 rc->warning->result,
2373 buffer_tostring(rc->warning->error_msg),
2377 warning_status = rrdcalc_value2status(rc->warning->result);
2381 if(likely(rc->critical)) {
2382 if(unlikely(!expression_evaluate(rc->critical))) {
2383 // calculation failed
2385 debug(D_HEALTH, "Health alarm '%s.%s': critical expression failed with error: %s",
2386 rc->chart?rc->chart:"NOCHART", rc->name, buffer_tostring(rc->critical->error_msg));
2388 if (unlikely(!(rc->rrdcalc_flags & RRDCALC_FLAG_CRIT_ERROR))) {
2389 rc->rrdcalc_flags |= RRDCALC_FLAG_CRIT_ERROR;
2390 error("Health alarm '%s.%s': critical expression failed with error: %s",
2391 rc->chart?rc->chart:"NOCHART", rc->name, buffer_tostring(rc->critical->error_msg));
2395 if(unlikely(rc->rrdcalc_flags & RRDCALC_FLAG_CRIT_ERROR))
2396 rc->rrdcalc_flags &= ~RRDCALC_FLAG_CRIT_ERROR;
2398 debug(D_HEALTH, "Health alarm '%s.%s': critical expression gave value "
2399 CALCULATED_NUMBER_FORMAT
2400 ": %s (source: %s)",
2401 rc->chart?rc->chart:"NOCHART", rc->name,
2402 rc->critical->result,
2403 buffer_tostring(rc->critical->error_msg),
2407 critical_status = rrdcalc_value2status(rc->critical->result);
2411 int status = RRDCALC_STATUS_UNDEFINED;
2413 switch(warning_status) {
2414 case RRDCALC_STATUS_CLEAR:
2415 status = RRDCALC_STATUS_CLEAR;
2418 case RRDCALC_STATUS_RAISED:
2419 status = RRDCALC_STATUS_WARNING;
2426 switch(critical_status) {
2427 case RRDCALC_STATUS_CLEAR:
2428 if(status == RRDCALC_STATUS_UNDEFINED)
2429 status = RRDCALC_STATUS_CLEAR;
2432 case RRDCALC_STATUS_RAISED:
2433 status = RRDCALC_STATUS_CRITICAL;
2440 if(status != rc->status) {
2443 if(now > rc->delay_up_to_timestamp) {
2444 rc->delay_up_current = rc->delay_up_duration;
2445 rc->delay_down_current = rc->delay_down_duration;
2447 rc->delay_up_to_timestamp = 0;
2450 rc->delay_up_current = (int)(rc->delay_up_current * rc->delay_multiplier);
2451 if(rc->delay_up_current > rc->delay_max_duration) rc->delay_up_current = rc->delay_max_duration;
2453 rc->delay_down_current = (int)(rc->delay_down_current * rc->delay_multiplier);
2454 if(rc->delay_down_current > rc->delay_max_duration) rc->delay_down_current = rc->delay_max_duration;
2457 if(status > rc->status)
2458 delay = rc->delay_up_current;
2460 delay = rc->delay_down_current;
2462 // COMMENTED: because we do need to send raising alarms
2463 // if(now + delay < rc->delay_up_to_timestamp)
2464 // delay = (int)(rc->delay_up_to_timestamp - now);
2466 rc->delay_last = delay;
2467 rc->delay_up_to_timestamp = now + delay;
2468 health_alarm_log(&localhost, rc->id, rc->next_event_id++, now, rc->name, rc->rrdset->id, rc->rrdset->family, rc->exec, rc->recipient, now - rc->last_status_change, rc->old_value, rc->value, rc->status, status, rc->source, rc->units, rc->info, rc->delay_last);
2469 rc->last_status_change = now;
2470 rc->status = status;
2473 rc->last_updated = now;
2474 rc->next_update = now + rc->update_every;
2476 if (next_run > rc->next_update)
2477 next_run = rc->next_update;
2480 rrdhost_unlock(&localhost);
2483 if (unlikely(pthread_setcancelstate(oldstate, NULL) != 0))
2484 error("Cannot set pthread cancel state to RESTORE (%d).", oldstate);
2486 // execute notifications
2488 health_alarm_log_process(&localhost);
2491 if(now < next_run) {
2492 debug(D_HEALTH, "Health monitoring iteration no %u done. Next iteration in %d secs",
2493 loop, (int) (next_run - now));
2494 sleep_usec(1000000 * (unsigned long long) (next_run - now));
2497 debug(D_HEALTH, "Health monitoring iteration no %u done. Next iteration now", loop);
2503 info("HEALTH thread exiting");