* {{{ */
#define UT_FLAG_INVERT 0x01
#define UT_FLAG_PERSIST 0x02
+#define UT_FLAG_PERCENTAGE 0x04
typedef struct threshold_s
{
return (-1);
}
memcpy (th_copy, th, sizeof (threshold_t));
- th_copy = NULL;
+ th_ptr = NULL;
DEBUG ("ut_threshold_add: Adding entry `%s'", name);
return (-1);
}
- strncpy (th->type_instance, ci->values[0].value.string,
+ sstrncpy (th->type_instance, ci->values[0].value.string,
sizeof (th->type_instance));
- th->type_instance[sizeof (th->type_instance) - 1] = '\0';
return (0);
} /* int ut_config_type_instance */
return (0);
} /* int ut_config_type_persist */
+static int ut_config_type_percentage(threshold_t *th, oconfig_item_t *ci)
+{
+ if ((ci->values_num != 1)
+ || (ci->values[0].type != OCONFIG_TYPE_BOOLEAN))
+ {
+ WARNING ("threshold values: The `Percentage' option needs exactly one "
+ "boolean argument.");
+ return (-1);
+ }
+
+ if (ci->values[0].value.boolean)
+ th->flags |= UT_FLAG_PERCENTAGE;
+ else
+ th->flags &= ~UT_FLAG_PERCENTAGE;
+
+ return (0);
+} /* int ut_config_type_percentage */
+
static int ut_config_type (const threshold_t *th_orig, oconfig_item_t *ci)
{
int i;
}
memcpy (&th, th_orig, sizeof (th));
- strncpy (th.type, ci->values[0].value.string, sizeof (th.type));
- th.type[sizeof (th.type) - 1] = '\0';
+ sstrncpy (th.type, ci->values[0].value.string, sizeof (th.type));
th.warning_min = NAN;
th.warning_max = NAN;
if (strcasecmp ("Instance", option->key) == 0)
status = ut_config_type_instance (&th, option);
- if (strcasecmp ("DataSource", option->key) == 0)
+ else if (strcasecmp ("DataSource", option->key) == 0)
status = ut_config_type_datasource (&th, option);
else if ((strcasecmp ("WarningMax", option->key) == 0)
|| (strcasecmp ("FailureMax", option->key) == 0))
status = ut_config_type_invert (&th, option);
else if (strcasecmp ("Persist", option->key) == 0)
status = ut_config_type_persist (&th, option);
+ else if (strcasecmp ("Percentage", option->key) == 0)
+ status = ut_config_type_percentage (&th, option);
else
{
WARNING ("threshold values: Option `%s' not allowed inside a `Type' "
return (-1);
}
- strncpy (th->plugin_instance, ci->values[0].value.string,
+ sstrncpy (th->plugin_instance, ci->values[0].value.string,
sizeof (th->plugin_instance));
- th->plugin_instance[sizeof (th->plugin_instance) - 1] = '\0';
return (0);
} /* int ut_config_plugin_instance */
}
memcpy (&th, th_orig, sizeof (th));
- strncpy (th.plugin, ci->values[0].value.string, sizeof (th.plugin));
- th.plugin[sizeof (th.plugin) - 1] = '\0';
+ sstrncpy (th.plugin, ci->values[0].value.string, sizeof (th.plugin));
for (i = 0; i < ci->children_num; i++)
{
}
memcpy (&th, th_orig, sizeof (th));
- strncpy (th.host, ci->values[0].value.string, sizeof (th.host));
- th.host[sizeof (th.host) - 1] = '\0';
+ sstrncpy (th.host, ci->values[0].value.string, sizeof (th.host));
for (i = 0; i < ci->children_num; i++)
{
th.warning_max = NAN;
th.failure_min = NAN;
th.failure_max = NAN;
-
+
for (i = 0; i < ci->children_num; i++)
{
oconfig_item_t *option = ci->children + i;
*/
/* }}} */
-static threshold_t *threshold_search (const data_set_t *ds,
- const value_list_t *vl)
+static threshold_t *threshold_search (const value_list_t *vl)
{
threshold_t *th;
if ((th = threshold_get (vl->host, vl->plugin, vl->plugin_instance,
- ds->type, vl->type_instance)) != NULL)
+ vl->type, vl->type_instance)) != NULL)
return (th);
else if ((th = threshold_get (vl->host, vl->plugin, vl->plugin_instance,
- ds->type, NULL)) != NULL)
+ vl->type, NULL)) != NULL)
return (th);
else if ((th = threshold_get (vl->host, vl->plugin, NULL,
- ds->type, vl->type_instance)) != NULL)
+ vl->type, vl->type_instance)) != NULL)
return (th);
else if ((th = threshold_get (vl->host, vl->plugin, NULL,
- ds->type, NULL)) != NULL)
+ vl->type, NULL)) != NULL)
return (th);
else if ((th = threshold_get (vl->host, "", NULL,
- ds->type, vl->type_instance)) != NULL)
+ vl->type, vl->type_instance)) != NULL)
return (th);
else if ((th = threshold_get (vl->host, "", NULL,
- ds->type, NULL)) != NULL)
+ vl->type, NULL)) != NULL)
return (th);
else if ((th = threshold_get ("", vl->plugin, vl->plugin_instance,
- ds->type, vl->type_instance)) != NULL)
+ vl->type, vl->type_instance)) != NULL)
return (th);
else if ((th = threshold_get ("", vl->plugin, vl->plugin_instance,
- ds->type, NULL)) != NULL)
+ vl->type, NULL)) != NULL)
return (th);
else if ((th = threshold_get ("", vl->plugin, NULL,
- ds->type, vl->type_instance)) != NULL)
+ vl->type, vl->type_instance)) != NULL)
return (th);
else if ((th = threshold_get ("", vl->plugin, NULL,
- ds->type, NULL)) != NULL)
+ vl->type, NULL)) != NULL)
return (th);
else if ((th = threshold_get ("", "", NULL,
- ds->type, vl->type_instance)) != NULL)
+ vl->type, vl->type_instance)) != NULL)
return (th);
else if ((th = threshold_get ("", "", NULL,
- ds->type, NULL)) != NULL)
+ vl->type, NULL)) != NULL)
return (th);
return (NULL);
} /* threshold_t *threshold_search */
-/* TODO: Split this into two functions, one that iterates over all threshold
- * definitions and one that checks if (a) the threshold matches (the data
- * source matches) and (b) if the threshold is satisfied or not. Break at the
- * first matching threshold. */
-int ut_check_threshold (const data_set_t *ds, const value_list_t *vl)
-{
+/*
+ * int ut_report_state
+ *
+ * Checks if the `state' differs from the old state and creates a notification
+ * if appropriate.
+ * Does not fail.
+ */
+static int ut_report_state (const data_set_t *ds,
+ const value_list_t *vl,
+ const threshold_t *th,
+ const gauge_t *values,
+ int ds_index,
+ int state)
+{ /* {{{ */
+ int state_old;
notification_t n;
- threshold_t *th;
- gauge_t *values;
- int i;
-
- int state_orig;
- int state_new = STATE_OKAY;
- int ds_index = 0;
char *buf;
size_t bufsize;
- int status;
-
- if (threshold_tree == NULL)
- return (0);
-
- /* Is this lock really necessary? So far, thresholds are only inserted at
- * startup. -octo */
- pthread_mutex_lock (&threshold_lock);
- th = threshold_search (ds, vl);
- pthread_mutex_unlock (&threshold_lock);
- if (th == NULL)
- return (0);
- DEBUG ("ut_check_threshold: Found matching threshold");
-
- values = uc_get_rate (ds, vl);
- if (values == NULL)
- return (0);
+ int status;
- state_orig = uc_get_state (ds, vl);
+ state_old = uc_get_state (ds, vl);
- for (i = 0; i < ds->ds_num; i++)
+ /* If the state didn't change, only report if `persistent' is specified and
+ * the state is not `okay'. */
+ if (state == state_old)
{
- int is_inverted = 0;
- int is_warning = 0;
- int is_failure = 0;
-
- if ((th->flags & UT_FLAG_INVERT) != 0)
- {
- is_inverted = 1;
- is_warning--;
- is_failure--;
- }
- if ((!isnan (th->failure_min) && (th->failure_min > values[i]))
- || (!isnan (th->failure_max) && (th->failure_max < values[i])))
- is_failure++;
- if ((!isnan (th->warning_min) && (th->warning_min > values[i]))
- || (!isnan (th->warning_max) && (th->warning_max < values[i])))
- is_warning++;
-
- if ((is_failure != 0) && (state_new != STATE_ERROR))
- {
- state_new = STATE_ERROR;
- ds_index = i;
- }
- else if ((is_warning != 0)
- && (state_new != STATE_ERROR)
- && (state_new != STATE_WARNING))
- {
- state_new = STATE_WARNING;
- ds_index = i;
- }
+ if ((th->flags & UT_FLAG_PERSIST) == 0)
+ return (0);
+ else if (state == STATE_OKAY)
+ return (0);
}
- if (state_new != state_orig)
- uc_set_state (ds, vl, state_new);
-
- /* Return here if we're not going to send a notification */
- if ((state_new == state_orig)
- && ((state_new == STATE_OKAY)
- || ((th->flags & UT_FLAG_PERSIST) == 0)))
- {
- sfree (values);
- return (0);
- }
+ if (state != state_old)
+ uc_set_state (ds, vl, state);
NOTIFICATION_INIT_VL (&n, vl, ds);
- {
- /* Copy the associative members */
- if (state_new == STATE_OKAY)
- n.severity = NOTIF_OKAY;
- else if (state_new == STATE_WARNING)
- n.severity = NOTIF_WARNING;
- else
- n.severity = NOTIF_FAILURE;
- n.time = vl->time;
+ buf = n.message;
+ bufsize = sizeof (n.message);
- buf = n.message;
- bufsize = sizeof (n.message);
+ if (state == STATE_OKAY)
+ n.severity = NOTIF_OKAY;
+ else if (state == STATE_WARNING)
+ n.severity = NOTIF_WARNING;
+ else
+ n.severity = NOTIF_FAILURE;
+
+ n.time = vl->time;
+
+ status = ssnprintf (buf, bufsize, "Host %s, plugin %s",
+ vl->host, vl->plugin);
+ buf += status;
+ bufsize -= status;
- status = snprintf (buf, bufsize, "Host %s, plugin %s",
- vl->host, vl->plugin);
+ if (vl->plugin_instance[0] != '\0')
+ {
+ status = ssnprintf (buf, bufsize, " (instance %s)",
+ vl->plugin_instance);
buf += status;
bufsize -= status;
+ }
- if (vl->plugin_instance[0] != '\0')
- {
- status = snprintf (buf, bufsize, " (instance %s)",
- vl->plugin_instance);
- buf += status;
- bufsize -= status;
- }
+ status = ssnprintf (buf, bufsize, " type %s", vl->type);
+ buf += status;
+ bufsize -= status;
- status = snprintf (buf, bufsize, " type %s", ds->type);
+ if (vl->type_instance[0] != '\0')
+ {
+ status = ssnprintf (buf, bufsize, " (instance %s)",
+ vl->type_instance);
buf += status;
bufsize -= status;
-
- if (vl->type_instance[0] != '\0')
- {
- status = snprintf (buf, bufsize, " (instance %s)",
- vl->type_instance);
- buf += status;
- bufsize -= status;
- }
}
- /* Send a okay notification */
- if (state_new == STATE_OKAY)
+ plugin_notification_meta_add_string (&n, "DataSource",
+ ds->ds[ds_index].name);
+ plugin_notification_meta_add_double (&n, "CurrentValue", values[ds_index]);
+ plugin_notification_meta_add_double (&n, "WarningMin", th->warning_min);
+ plugin_notification_meta_add_double (&n, "WarningMax", th->warning_max);
+ plugin_notification_meta_add_double (&n, "FailureMin", th->failure_min);
+ plugin_notification_meta_add_double (&n, "FailureMax", th->failure_max);
+
+ /* Send an okay notification */
+ if (state == STATE_OKAY)
{
- status = snprintf (buf, bufsize, ": All data sources are within range again.");
+ status = ssnprintf (buf, bufsize, ": All data sources are within range again.");
buf += status;
bufsize -= status;
}
double min;
double max;
- min = (state_new == STATE_ERROR) ? th->failure_min : th->warning_min;
- max = (state_new == STATE_ERROR) ? th->failure_max : th->warning_max;
+ min = (state == STATE_ERROR) ? th->failure_min : th->warning_min;
+ max = (state == STATE_ERROR) ? th->failure_max : th->warning_max;
if (th->flags & UT_FLAG_INVERT)
{
if (!isnan (min) && !isnan (max))
{
- status = snprintf (buf, bufsize, ": Data source \"%s\" is currently "
- "%f. That is within the %s region of %f and %f.",
- ds->ds[ds_index].name, values[ds_index],
- (state_new == STATE_ERROR) ? "failure" : "warning",
- min, min);
+ status = ssnprintf (buf, bufsize, ": Data source \"%s\" is currently "
+ "%f. That is within the %s region of %f%s and %f%s.",
+ ds->ds[ds_index].name, values[ds_index],
+ (state == STATE_ERROR) ? "failure" : "warning",
+ min, ((th->flags & UT_FLAG_PERCENTAGE) != 0) ? "%" : "",
+ max, ((th->flags & UT_FLAG_PERCENTAGE) != 0) ? "%" : "");
}
else
{
- status = snprintf (buf, bufsize, ": Data source \"%s\" is currently "
- "%f. That is %s the %s threshold of %f.",
+ status = ssnprintf (buf, bufsize, ": Data source \"%s\" is currently "
+ "%f. That is %s the %s threshold of %f%s.",
ds->ds[ds_index].name, values[ds_index],
isnan (min) ? "below" : "above",
- (state_new == STATE_ERROR) ? "failure" : "warning",
- isnan (min) ? max : min);
+ (state == STATE_ERROR) ? "failure" : "warning",
+ isnan (min) ? max : min,
+ ((th->flags & UT_FLAG_PERCENTAGE) != 0) ? "%" : "");
}
}
else /* is not inverted */
{
- status = snprintf (buf, bufsize, ": Data source \"%s\" is currently "
- "%f. That is %s the %s threshold of %f.",
+ status = ssnprintf (buf, bufsize, ": Data source \"%s\" is currently "
+ "%f. That is %s the %s threshold of %f%s.",
ds->ds[ds_index].name, values[ds_index],
(values[ds_index] < min) ? "below" : "above",
- (state_new == STATE_ERROR) ? "failure" : "warning",
- (values[ds_index] < min) ? min : max);
+ (state == STATE_ERROR) ? "failure" : "warning",
+ (values[ds_index] < min) ? min : max,
+ ((th->flags & UT_FLAG_PERCENTAGE) != 0) ? "%" : "");
}
buf += status;
bufsize -= status;
plugin_dispatch_notification (&n);
+ plugin_notification_meta_free (n.meta);
+ return (0);
+} /* }}} int ut_report_state */
+
+/*
+ * int ut_check_one_data_source
+ *
+ * Checks one data source against the given threshold configuration. If the
+ * `DataSource' option is set in the threshold, and the name does NOT match,
+ * `okay' is returned. If the threshold does match, its failure and warning
+ * min and max values are checked and `failure' or `warning' is returned if
+ * appropriate.
+ * Does not fail.
+ */
+static int ut_check_one_data_source (const data_set_t *ds,
+ const value_list_t __attribute__((unused)) *vl,
+ const threshold_t *th,
+ const gauge_t *values,
+ int ds_index)
+{ /* {{{ */
+ const char *ds_name;
+ int is_warning = 0;
+ int is_failure = 0;
+
+ /* check if this threshold applies to this data source */
+ if (ds != NULL)
+ {
+ ds_name = ds->ds[ds_index].name;
+ if ((th->data_source[0] != 0)
+ && (strcmp (ds_name, th->data_source) != 0))
+ return (STATE_OKAY);
+ }
+
+ if ((th->flags & UT_FLAG_INVERT) != 0)
+ {
+ is_warning--;
+ is_failure--;
+ }
+
+ if ((!isnan (th->failure_min) && (th->failure_min > values[ds_index]))
+ || (!isnan (th->failure_max) && (th->failure_max < values[ds_index])))
+ is_failure++;
+ if (is_failure != 0)
+ return (STATE_ERROR);
+
+ 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_warning != 0)
+ return (STATE_WARNING);
+
+ return (STATE_OKAY);
+} /* }}} int ut_check_one_data_source */
+
+/*
+ * int ut_check_one_threshold
+ *
+ * Checks all data sources of a value list against the given threshold, using
+ * the ut_check_one_data_source function above. Returns the worst status,
+ * which is `okay' if nothing has failed.
+ * Returns less than zero if the data set doesn't have any data sources.
+ */
+static int ut_check_one_threshold (const data_set_t *ds,
+ const value_list_t *vl,
+ const threshold_t *th,
+ const gauge_t *values,
+ int *ret_ds_index)
+{ /* {{{ */
+ int ret = -1;
+ int ds_index = -1;
+ int i;
+ gauge_t values_copy[ds->ds_num];
+
+ memcpy (values_copy, values, sizeof (values_copy));
+
+ if ((th->flags & UT_FLAG_PERCENTAGE) != 0)
+ {
+ int num = 0;
+ gauge_t sum=0.0;
+
+ if (ds->ds_num == 1)
+ {
+ WARNING ("ut_check_one_threshold: The %s type has only one data "
+ "source, but you have configured to check this as a percentage. "
+ "That doesn't make much sense, because the percentage will always "
+ "be 100%%!", ds->type);
+ }
+
+ /* Prepare `sum' and `num'. */
+ for (i = 0; i < ds->ds_num; i++)
+ if (!isnan (values[i]))
+ {
+ num++;
+ sum += values[i];
+ }
+
+ if ((num == 0) /* All data sources are undefined. */
+ || (sum == 0.0)) /* Sum is zero, cannot calculate percentage. */
+ {
+ for (i = 0; i < ds->ds_num; i++)
+ values_copy[i] = NAN;
+ }
+ else /* We can actually calculate the percentage. */
+ {
+ for (i = 0; i < ds->ds_num; i++)
+ values_copy[i] = 100.0 * values[i] / sum;
+ }
+ } /* if (UT_FLAG_PERCENTAGE) */
+
+ for (i = 0; i < ds->ds_num; i++)
+ {
+ int status;
+
+ status = ut_check_one_data_source (ds, vl, th, values_copy, i);
+ if (ret < status)
+ {
+ ret = status;
+ ds_index = i;
+ }
+ } /* for (ds->ds_num) */
+
+ if (ret_ds_index != NULL)
+ *ret_ds_index = ds_index;
+
+ return (ret);
+} /* }}} int ut_check_one_threshold */
+
+/*
+ * int ut_check_threshold (PUBLIC)
+ *
+ * Gets a list of matching thresholds and searches for the worst status by one
+ * of the thresholds. Then reports that status using the ut_report_state
+ * function above.
+ * Returns zero on success and if no threshold has been configured. Returns
+ * less than zero on failure.
+ */
+int ut_check_threshold (const data_set_t *ds, const value_list_t *vl)
+{ /* {{{ */
+ threshold_t *th;
+ gauge_t *values;
+ int status;
+
+ int worst_state = -1;
+ threshold_t *worst_th = NULL;
+ int worst_ds_index = -1;
+
+ if (threshold_tree == NULL)
+ return (0);
+
+ /* Is this lock really necessary? So far, thresholds are only inserted at
+ * startup. -octo */
+ pthread_mutex_lock (&threshold_lock);
+ th = threshold_search (vl);
+ pthread_mutex_unlock (&threshold_lock);
+ if (th == NULL)
+ return (0);
+
+ DEBUG ("ut_check_threshold: Found matching threshold(s)");
+
+ values = uc_get_rate (ds, vl);
+ if (values == NULL)
+ return (0);
+
+ while (th != NULL)
+ {
+ int ds_index = -1;
+
+ status = ut_check_one_threshold (ds, vl, th, values, &ds_index);
+ if (status < 0)
+ {
+ ERROR ("ut_check_threshold: ut_check_one_threshold failed.");
+ sfree (values);
+ return (-1);
+ }
+
+ if (worst_state < status)
+ {
+ worst_state = status;
+ worst_th = th;
+ worst_ds_index = ds_index;
+ }
+
+ th = th->next;
+ } /* while (th) */
+
+ status = ut_report_state (ds, vl, worst_th, values,
+ worst_ds_index, worst_state);
+ if (status != 0)
+ {
+ ERROR ("ut_check_threshold: ut_report_state failed.");
+ sfree (values);
+ return (-1);
+ }
+
sfree (values);
return (0);
-} /* int ut_check_threshold */
+} /* }}} int ut_check_threshold */
+/*
+ * int ut_check_interesting (PUBLIC)
+ *
+ * Given an identification returns
+ * 0: No threshold is defined.
+ * 1: A threshold has been found. The flag `persist' is off.
+ * 2: A threshold has been found. The flag `persist' is on.
+ * (That is, it is expected that many notifications are sent until the
+ * problem disappears.)
+ */
int ut_check_interesting (const char *name)
-{
+{ /* {{{ */
char *name_copy = NULL;
char *host = NULL;
char *plugin = NULL;
if (status != 0)
{
ERROR ("ut_check_interesting: parse_identifier failed.");
+ sfree (name_copy);
return (-1);
}
memset (&ds, '\0', sizeof (ds));
memset (&vl, '\0', sizeof (vl));
- strncpy (vl.host, host, sizeof (vl.host));
- vl.host[sizeof (vl.host) - 1] = '\0';
- strncpy (vl.plugin, plugin, sizeof (vl.plugin));
- vl.plugin[sizeof (vl.plugin) - 1] = '\0';
+ sstrncpy (vl.host, host, sizeof (vl.host));
+ sstrncpy (vl.plugin, plugin, sizeof (vl.plugin));
if (plugin_instance != NULL)
- {
- strncpy (vl.plugin_instance, plugin_instance, sizeof (vl.plugin_instance));
- vl.plugin_instance[sizeof (vl.plugin_instance) - 1] = '\0';
- }
- strncpy (ds.type, type, sizeof (ds.type));
- ds.type[sizeof (ds.type) - 1] = '\0';
+ sstrncpy (vl.plugin_instance, plugin_instance, sizeof (vl.plugin_instance));
+ sstrncpy (ds.type, type, sizeof (ds.type));
+ sstrncpy (vl.type, type, sizeof (vl.type));
if (type_instance != NULL)
- {
- strncpy (vl.type_instance, type_instance, sizeof (vl.type_instance));
- vl.type_instance[sizeof (vl.type_instance) - 1] = '\0';
- }
+ sstrncpy (vl.type_instance, type_instance, sizeof (vl.type_instance));
sfree (name_copy);
host = plugin = plugin_instance = type = type_instance = NULL;
- th = threshold_search (&ds, &vl);
+ th = threshold_search (&vl);
if (th == NULL)
return (0);
if ((th->flags & UT_FLAG_PERSIST) == 0)
return (1);
return (2);
-} /* int ut_check_interesting */
+} /* }}} int ut_check_interesting */
-/* vim: set sw=2 ts=8 sts=2 tw=78 fdm=marker : */
+/* vim: set sw=2 ts=8 sts=2 tw=78 et fdm=marker : */
projects / collectd.git / blobdiff