Merge branch 'collectd-5.4' into collectd-5.5

[collectd.git] / src / threshold.c

diff --git a/src/threshold.c b/src/threshold.c
index 855f136..8815a00 100644 (file)
--- a/src/threshold.c
+++ b/src/threshold.c
@@ -519,7 +519,9 @@ static int ut_report_state (const data_set_t *ds,
           ": Value is no longer missing.");
     else
       status = ssnprintf (buf, bufsize,
-          ": All data sources are within range again.");
+          ": All data sources are within range again. "
+          "Current value of \"%s\" is %f.",
+          ds->ds[ds_index].name, values[ds_index]);
     buf += status;
     bufsize -= status;
   }
@@ -637,23 +639,40 @@ static int ut_check_one_data_source (const data_set_t *ds,
 
   /* XXX: This is an experimental code, not optimized, not fast, not reliable,
    * and probably, do not work as you expect. Enjoy! :D */
-  if ( (th->hysteresis > 0) && ((prev_state = uc_get_state(ds,vl)) != STATE_OKAY) )
-  {
-    switch(prev_state)
+  if (th->hysteresis > 0)
+  {
+    prev_state = uc_get_state(ds,vl);
+    /* The purpose of hysteresis is elliminating flapping state when the value
+     * oscilates around the thresholds. In other words, what is important is
+     * the previous state; if the new value would trigger a transition, make
+     * sure that we artificially widen the range which is considered to apply
+     * for the previous state, and only trigger the notification if the value
+     * is outside of this expanded range.
+     *
+     * There is no hysteresis for the OKAY state.
+     * */
+    gauge_t hysteresis_for_warning = 0, hysteresis_for_failure = 0;
+    switch (prev_state)
     {
       case STATE_ERROR:
-       if ( (!isnan (th->failure_min) && ((th->failure_min + th->hysteresis) < values[ds_index])) ||
-            (!isnan (th->failure_max) && ((th->failure_max - th->hysteresis) > values[ds_index])) )
-         return (STATE_OKAY);
-       else
-         is_failure++;
+        hysteresis_for_failure = th->hysteresis;
+        break;
       case STATE_WARNING:
-       if ( (!isnan (th->warning_min) && ((th->warning_min + th->hysteresis) < values[ds_index])) ||
-            (!isnan (th->warning_max) && ((th->warning_max - th->hysteresis) > values[ds_index])) )
-         return (STATE_OKAY);
-       else
-         is_warning++;
-     }
+        hysteresis_for_warning = th->hysteresis;
+        break;
+      case STATE_OKAY:
+        /* do nothing -- the hysteresis only applies to the non-normal states */
+        break;
+    }
+
+    if ((!isnan (th->failure_min) && (th->failure_min + hysteresis_for_failure > values[ds_index]))
+       || (!isnan (th->failure_max) && (th->failure_max - hysteresis_for_failure < values[ds_index])))
+      is_failure++;
+
+    if ((!isnan (th->warning_min) && (th->warning_min + hysteresis_for_warning > values[ds_index]))
+       || (!isnan (th->warning_max) && (th->warning_max - hysteresis_for_warning < values[ds_index])))
+      is_warning++;
+
   }
   else { /* no hysteresis */
     if ((!isnan (th->failure_min) && (th->failure_min > values[ds_index]))
@@ -663,7 +682,7 @@ static int ut_check_one_data_source (const data_set_t *ds,
     if ((!isnan (th->warning_min) && (th->warning_min > values[ds_index]))
        || (!isnan (th->warning_max) && (th->warning_max < values[ds_index])))
       is_warning++;
- }
+  }
 
   if (is_failure != 0)
     return (STATE_ERROR);
@@ -832,6 +851,7 @@ static int ut_missing (const value_list_t *vl,
   cdtime_t missing_time;
   char identifier[6 * DATA_MAX_NAME_LEN];
   notification_t n;
+  cdtime_t now;
 
   if (threshold_tree == NULL)
     return (0);
@@ -841,13 +861,15 @@ static int ut_missing (const value_list_t *vl,
   if ((th == NULL) || ((th->flags & UT_FLAG_INTERESTING) == 0))
     return (0);
 
-  missing_time = cdtime () - vl->time;
+  now = cdtime ();
+  missing_time = now - vl->time;
   FORMAT_VL (identifier, sizeof (identifier), vl);
 
   NOTIFICATION_INIT_VL (&n, vl);
   ssnprintf (n.message, sizeof (n.message),
       "%s has not been updated for %.3f seconds.",
       identifier, CDTIME_T_TO_DOUBLE (missing_time));
+  n.time = now;
 
   plugin_dispatch_notification (&n);