The following are parameters that are common to more than one aggregator.
unitUnit is a time unit represented as a string and must be one of ( MILLISECONDS, SECONDS, MINUTES, HOURS, DAYS, WEEKS, MONTHS, YEARS)
samplingA sampling is a json object containing two values a value and a unit.
The value is a long and the unit is see unit.
Many of the below aggregators inherit from the range aggregator. You can set the following parameters on any range aggregator.
sampling (Sampling {value (long), unit (See unit)})
Sampling is the length of the interval on which to aggregate data.
"aggregators": [
{
"name": "sum",
"align_sampling": true,
"align_start_time": true,
"align_end_time": false,
"sampling": {
"value": 1,
"unit": "minutes"
}
}]
align_start_time - (boolean, optional, default value: false)
When set to true the time for the aggregated data point for each range will fall on the start of the range instead of being the value for the first data point within that range. Note that align_sampling, align_start_time, and align_end_time are mutually exclusive. If more than one are set, unexpected results will occur.
align_end_time - (boolean, optional, default value: false)
Setting this to true will cause the aggregation range to be aligned based on the sampling size. For example if your sample size is either milliseconds, seconds, minutes or hours then the start of the range will always be at the top of the hour. The difference between align_start_time and align_end_time is that align_end_time sets the timestamp for the datapoint to the beginning of the following period versus the beginning of the current period. As with align_start_time, setting this to true will cause your data to take the same shape when graphed as you refresh the data. Note that align_start_time and align_end_time are mutually exclusive. If more than one are set, unexpected results will occur.
align_sampling - (boolean, optional, default value: false)
Setting this to true will cause the aggregation range to be aligned based on the sampling size. For example if your sample size is either milliseconds, seconds, minutes or hours then the start of the range will always be at the top of the hour. The effect of setting this to true is that your data will take the same shape when graphed as you refresh the data. Note that align_sampling, align_start_time, and align_end_time are mutually exclusive. If more than one are set, unexpected results will occur.
start_time - (long, optional, default value: 0)
Start time to calculate the ranges from. Typically this is the start of the query
time_zone - (long time zone format)
Time zone to use when doing time based calculations.
avg¶Computes average value. Extends Range Aggregator.
dev¶Computes standard deviation. Extends Range Aggregator.
countCounts the number of data points. Extends Range Aggregator.
firstReturns the first data point for the interval. Extends Range Aggregator.
gapsMarks gaps in data according to sampling rate with a null data point. Extends Range Aggregator.
histogramCalculates a probability distribution and returns the specified percentile for the distribution. The “percentile” value is defined as 0 < percentile <= 1 where .5 is 50% and 1 is 100%. Note that this aggregator has been renamed to percentile in release 0.9.2. See Percentile.
lastReturns the last data point for the interval. Extends Range Aggregator.
least_squares¶Returns two points for the range which represent the best fit line through the set of points. Extends Range Aggregator.
maxReturns the largest value in the interval. Extends Range Aggregator.
minReturns the smallest value in the interval. Extends Range Aggregator.
percentileFinds the percentile of the data range. Calculates a probability distribution and returns the specified percentile for the distribution. The “percentile” value is defined as 0 < percentile <= 1 where .5 is 50% and 1 is 100%. Extends Range Aggregator.
sumSums all values Extends Range Aggregator.
diffComputes the difference between successive data points.
div¶Returns each data point divided by a divisor. Requires a “divisor” property which is the value that all data points will be divided by.
rateReturns the rate of change between a pair of data points. Requires a “unit” property which is the sampling duration (ie rate in seconds, milliseconds, minutes, etc…).
sampling (See sampling) - Sets the sampling for calculating the rate.
unit (See unit) - Shortcut for setting the sampling to a single unit.
If you set the unit to SECONDS then the sampling is over one second.
time_zone (Long format time zone) - Time zone for doing time calculations.
samplerComputes the sampling rate of change for the data points. Requires a “unit” property which is the sampling duration (ie rate in seconds, milliseconds, minutes, etc…).
unit (See unit) - Sets the sampling unit.
If you set the unit to SECONDS then the sampling rate is over one second.
time_zone (Long format time zone) - Time zone for doing time calculations.
scaleScales each data point by a factor. Requires a “factor” property which is the scaling value.
trimTrims off the first, last or both data points for the interval. Useful in conjunction with the save_as aggregator to remove partial intervals.
save_as¶Saves the result to another metric. Any data point with a unique tag value will also
have that tag set. So if a data point is returned with tags {"dc":["DC1"],"host":["hostA", "hostB"]}
only the dc tag will be set when saved. If you do a group by query the group by tags are saved.
metric_name (string) - Metric name to save the results to.
tags (Map of key values) - Additional tags to set on the metrics {"tag1":"value1","tag2":"value2"}
ttl (integer) - Sets the ttl on the newly saved metrics
add_saved_from (boolean) - Tells the aggregator to add the saved_from tag to the new metric. Defaults to true.
filterFilters out data points matching given critera.
filter_op (LTE, LT, GTE, GT, EQUAL) - Defines what data points to filter in relation to the threshold.
threshold (double) - Sets the threshold value for filtering data points.
scoreScores the data based on a set of thresholds. Each data point will be mapped to a value between 0 and n where n is the number of thresholds.
order (ASCENDING, DESCENDING) - The order by which scores are assigned.
thresholds (list) - A set of thresholds to compare the data against, where a threshold is an object with the following properties:
js_function¶js_filter¶js_range¶The JS Aggregator is provided as a third party module found here
https://github.com/Kratos-ISE/kise-kairosdb-module/
The module requires Java 8 and provides a way to pass javascript code as the aggregator.
Note. this project is now out of date with current version of Kairos.