1 #ifndef NETDATA_HEALTH_H
2 #define NETDATA_HEALTH_H
4 extern int health_enabled;
6 extern int rrdvar_compare(void *a, void *b);
8 #define RRDVAR_TYPE_CALCULATED 1
9 #define RRDVAR_TYPE_TIME_T 2
10 #define RRDVAR_TYPE_COLLECTED 3
11 #define RRDVAR_TYPE_TOTAL 4
12 #define RRDVAR_TYPE_INT 5
14 // the variables as stored in the variables indexes
15 // there are 3 indexes:
16 // 1. at each chart (RRDSET.variables_root_index)
17 // 2. at each context (RRDFAMILY.variables_root_index)
18 // 3. at each host (RRDHOST.variables_root_index)
19 typedef struct rrdvar {
31 // variables linked to charts
32 // We link variables to point to the values that are already
33 // calculated / processed by the normal data collection process
34 // This means, there will be no speed penalty for using
36 typedef struct rrdsetvar {
37 char *fullid; // chart type.chart id.variable
38 char *fullname; // chart type.chart name.variable
39 char *variable; // variable
52 struct rrdset *rrdset;
54 struct rrdsetvar *next;
58 // variables linked to individual dimensions
59 // We link variables to point the values that are already
60 // calculated / processed by the normal data collection process
61 // This means, there will be no speed penalty for using
63 typedef struct rrddimvar {
67 char *id; // dimension id
68 char *name; // dimension name
69 char *fullidid; // chart type.chart id.dimension id
70 char *fullidname; // chart type.chart id.dimension name
71 char *fullnameid; // chart type.chart name.dimension id
72 char *fullnamename; // chart type.chart name.dimension name
85 RRDVAR *host_fullidid;
86 RRDVAR *host_fullidname;
87 RRDVAR *host_fullnameid;
88 RRDVAR *host_fullnamename;
90 struct rrddim *rrddim;
92 struct rrddimvar *next;
95 // calculated variables (defined in health configuration)
96 // These aggregate time-series data at fixed intervals
97 // (defined in their update_every member below)
98 // These increase the overhead of netdata.
100 // These calculations are allocated and linked (->next)
102 // Then are also linked to RRDSET (of course only when the
103 // chart is found, via ->rrdset_next and ->rrdset_prev).
104 // This double-linked list is maintained sorted at all times
105 // having as RRDSET.calculations the RRDCALC to be processed
108 #define RRDCALC_STATUS_REMOVED -2
109 #define RRDCALC_STATUS_UNDEFINED -1
110 #define RRDCALC_STATUS_UNINITIALIZED 0
111 #define RRDCALC_STATUS_CLEAR 1
112 #define RRDCALC_STATUS_RAISED 2
113 #define RRDCALC_STATUS_WARNING 3
114 #define RRDCALC_STATUS_CRITICAL 4
116 #define RRDCALC_FLAG_DB_ERROR 0x00000001
117 #define RRDCALC_FLAG_DB_NAN 0x00000002
118 /* #define RRDCALC_FLAG_DB_STALE 0x00000004 */
119 #define RRDCALC_FLAG_CALC_ERROR 0x00000008
120 #define RRDCALC_FLAG_WARN_ERROR 0x00000010
121 #define RRDCALC_FLAG_CRIT_ERROR 0x00000020
122 #define RRDCALC_FLAG_RUNNABLE 0x00000040
124 typedef struct rrdcalc {
125 uint32_t id; // the unique id of this alarm
126 uint32_t next_event_id; // the next event id that will be used for this alarm
128 char *name; // the name of this alarm
131 char *exec; // the command to execute when this alarm switches state
132 char *recipient; // the recipient of the alarm (the first parameter to exec)
134 char *chart; // the chart id this should be linked to
137 char *source; // the source of this alarm
138 char *units; // the units of the alarm
139 char *info; // a short description of the alarm
141 int update_every; // update frequency for the alarm
143 // the red and green threshold of this alarm (to be set to the chart)
144 calculated_number green;
145 calculated_number red;
147 // ------------------------------------------------------------------------
148 // database lookup settings
150 char *dimensions; // the chart dimensions
151 int group; // grouping method: average, max, etc.
152 int before; // ending point in time-series
153 int after; // starting point in time-series
154 uint32_t options; // calculation options
156 // ------------------------------------------------------------------------
157 // expressions related to the alarm
159 EVAL_EXPRESSION *calculation; // expression to calculate the value of the alarm
160 EVAL_EXPRESSION *warning; // expression to check the warning condition
161 EVAL_EXPRESSION *critical; // expression to check the critical condition
163 // ------------------------------------------------------------------------
164 // notification delay settings
166 int delay_up_duration; // duration to delay notifications when alarm raises
167 int delay_down_duration; // duration to delay notifications when alarm lowers
168 int delay_max_duration; // the absolute max delay to apply to this alarm
169 float delay_multiplier; // multiplier for all delays when alarms switch status
170 // while now < delay_up_to
172 // ------------------------------------------------------------------------
173 // runtime information
175 int status; // the current status of the alarm
177 calculated_number value; // the current value of the alarm
178 calculated_number old_value; // the previous value of the alarm
180 uint32_t rrdcalc_flags; // check RRDCALC_FLAG_*
182 time_t last_updated; // the last update timestamp of the alarm
183 time_t next_update; // the next update timestamp of the alarm
184 time_t last_status_change; // the timestamp of the last time this alarm changed status
186 time_t db_after; // the first timestamp evaluated by the db lookup
187 time_t db_before; // the last timestamp evaluated by the db lookup
189 time_t delay_up_to_timestamp; // the timestamp up to which we should delay notifications
190 int delay_up_current; // the current up notification delay duration
191 int delay_down_current; // the current down notification delay duration
192 int delay_last; // the last delay we used
194 // ------------------------------------------------------------------------
195 // variables this alarm exposes to the rest of the alarms
202 // ------------------------------------------------------------------------
203 // the chart this alarm it is linked to
205 struct rrdset *rrdset;
207 // linking of this alarm on its chart
208 struct rrdcalc *rrdset_next;
209 struct rrdcalc *rrdset_prev;
211 struct rrdcalc *next;
214 #define RRDCALC_HAS_DB_LOOKUP(rc) ((rc)->after)
217 // these are to be applied to charts found dynamically
218 // based on their context.
219 typedef struct rrdcalctemplate {
227 uint32_t hash_context;
229 char *source; // the source of this alarm
230 char *units; // the units of the alarm
231 char *info; // a short description of the alarm
233 int update_every; // update frequency for the alarm
235 // the red and green threshold of this alarm (to be set to the chart)
236 calculated_number green;
237 calculated_number red;
239 // ------------------------------------------------------------------------
240 // database lookup settings
242 char *dimensions; // the chart dimensions
243 int group; // grouping method: average, max, etc.
244 int before; // ending point in time-series
245 int after; // starting point in time-series
246 uint32_t options; // calculation options
248 // ------------------------------------------------------------------------
249 // notification delay settings
251 int delay_up_duration; // duration to delay notifications when alarm raises
252 int delay_down_duration; // duration to delay notifications when alarm lowers
253 int delay_max_duration; // the absolute max delay to apply to this alarm
254 float delay_multiplier; // multiplier for all delays when alarms switch status
256 // ------------------------------------------------------------------------
257 // expressions related to the alarm
259 EVAL_EXPRESSION *calculation;
260 EVAL_EXPRESSION *warning;
261 EVAL_EXPRESSION *critical;
263 struct rrdcalctemplate *next;
266 #define RRDCALCTEMPLATE_HAS_CALCULATION(rt) ((rt)->after)
268 #define HEALTH_ENTRY_FLAG_PROCESSED 0x00000001
269 #define HEALTH_ENTRY_FLAG_UPDATED 0x00000002
270 #define HEALTH_ENTRY_FLAG_EXEC_RUN 0x00000004
271 #define HEALTH_ENTRY_FLAG_EXEC_FAILED 0x00000008
272 #define HEALTH_ENTRY_FLAG_SAVED 0x10000000
274 typedef struct alarm_entry {
277 uint32_t alarm_event_id;
281 time_t non_clear_duration;
293 time_t exec_run_timestamp;
300 calculated_number old_value;
301 calculated_number new_value;
308 time_t delay_up_to_timestamp;
310 uint32_t updated_by_id;
313 struct alarm_entry *next;
316 typedef struct alarm_log {
317 uint32_t next_log_id;
318 uint32_t next_alarm_id;
322 pthread_rwlock_t alarm_log_rwlock;
327 extern void rrdsetvar_rename_all(RRDSET *st);
328 extern RRDSETVAR *rrdsetvar_create(RRDSET *st, const char *variable, int type, void *value, uint32_t options);
329 extern void rrdsetvar_free(RRDSETVAR *rs);
331 extern void rrddimvar_rename_all(RRDDIM *rd);
332 extern RRDDIMVAR *rrddimvar_create(RRDDIM *rd, int type, const char *prefix, const char *suffix, void *value, uint32_t options);
333 extern void rrddimvar_free(RRDDIMVAR *rs);
335 extern void rrdsetcalc_link_matching(RRDSET *st);
336 extern void rrdsetcalc_unlink(RRDCALC *rc);
337 extern void rrdcalctemplate_link_matching(RRDSET *st);
338 extern RRDCALC *rrdcalc_find(RRDSET *st, const char *name);
340 extern void health_init(void);
341 extern void *health_main(void *ptr);
343 extern void health_reload(void);
345 extern int health_variable_lookup(const char *variable, uint32_t hash, RRDCALC *rc, calculated_number *result);
346 extern void health_alarms2json(RRDHOST *host, BUFFER *wb, int all);
347 extern void health_alarm_log2json(RRDHOST *host, BUFFER *wb, uint32_t after);
349 #endif //NETDATA_HEALTH_H