Table of Contents

FIXME This page is not fully translated, yet. Please help completing the translation.
(remove this paragraph once the translation is finished)

7 Prediction functions

All functions listed here are supported in:

Some general notes on function parameters:

  • Function parameters are separated by a comma
  • Optional function parameters (or parameter parts) are indicated by < >
  • /host/key is a common mandatory first parameter for the functions referencing the host item history
  • /host/key and time period:time shift parameters must never be quoted
  • Function-specific parameters are described with each function
FUNCTION
Description Function-specific parameters Comments
forecast (/host/key,time period<:time shift>,time,<fit>,<mode>)
Future value, max, min, delta or avg of the item. time period - maximum evaluation period1 in seconds (time suffixes can be used) or in latest collected values (if preceded by a hash mark)

Time shift (optional) allows to move the evaluation point back in time.

time - forecasting horizon in seconds (time suffixes can be used); negative values are supported

fit (optional; must be double-quoted) - function used to fit historical data

Supported fits:
linear - linear function
polynomialN - polynomial of degree N (1 <= N <= 6)
exponential - exponential function
logarithmic - logarithmic function
power - power function

Note that:
linear is default, polynomial1 is equivalent to linear

mode (optional; must be double-quoted) - demanded output

Supported modes:
value - value (default)
max - maximum
min - minimum
delta - max-min
avg - average

Note that:
value estimates item value at the moment now + time
max, min, delta and avg investigate item value estimate on the interval between now and now + time
Supported value types: float, int

If value to return is larger than 1.7976931348623157E+308 or less than -1.7976931348623157E+308, return value is cropped to 1.7976931348623157E+308 or -1.7976931348623157E+308 correspondingly.

Becomes unsupported only if misused in expression (wrong item type, invalid parameters), otherwise returns -1 in case of errors.

Examples:
=> forecast(/host/key,#10,1h) → forecast item value in one hour based on the last 10 values
=> forecast(/host/key,1h,30m) → forecast item value in 30 minutes based on the last hour data1
=> forecast(/host/key,1h:now-1d,12h) → forecast item value in 12 hours based on one hour one day ago
=> forecast(/host/key,1h,10m,"exponential") → forecast item value in 10 minutes based on the last hour data and exponential function
=> forecast(/host/key,1h,2h,"polynomial3","max") → forecast the maximum value the item can reach in the next two hours based on last hour data and cubic (third degree) polynomial
=> forecast(/host/key,#2,-20m) → estimate the item value 20 minutes ago based on the last two values (this can be more precise than using last(), especially if item is updated rarely, say, once an hour)

See also additional information on predictive trigger functions.
timeleft (/host/key,time period<:time shift>,threshold,<fit>)
Time in seconds needed for an item to reach a specified threshold. time period - maximum evaluation period1 in seconds (time suffixes can be used) or in latest collected values (if preceded by a hash mark)

Time shift (optional) allows to move the evaluation point back in time.

threshold - value to reach (unit suffixes can be used)

fit (optional; must be double-quoted) - see forecast()
Supported value types: float, int

If value to return is larger than 1.7976931348623157E+308, return value is cropped to 1.7976931348623157E+308.

Returns 1.7976931348623157E+308 if threshold cannot be reached.

Becomes unsupported only if misused in the expression (wrong item type, invalid parameters), otherwise returns -1 in case of errors.

Examples:
=> timeleft(/host/key,#10,0) → time until the item value reaches zero based on the last 10 values
=> timeleft(/host/key,1h,100) → time until the item value reaches 100 based on the last hour data
=> timeleft(/host/key,1h:now-1d,100) → time until the item value reaches 100 based on one hour one day ago
=> timeleft(/host/key,1h,200,"polynomial2") → time until the item value reaches 200 based on the last hour data and assumption that the item behaves like quadratic (second degree) polynomial
See also additional information on predictive trigger functions.
Footnotes

1 The function is evaluated starting with the first received value (unless the time shift parameter is used), i.e. Zabbix will not wait for all the values of the defined evaluation period before evaluating the function.