Operate on timestamps with Flux

This page documents an earlier version of InfluxDB. InfluxDB v2.7 is the latest stable version. View this page in the v2.7 documentation.

Every point stored in InfluxDB has an associated timestamp. Use Flux to process and operate on timestamps to suit your needs.

If you’re just getting started with Flux queries, check out the following:

Convert timestamp format

Unix nanosecond to RFC3339

Use the time() function to convert a Unix nanosecond timestamp to an RFC3339 timestamp.

  1. time(v: 1568808000000000000)
  2. // Returns 2019-09-18T12:00:00.000000000Z

RFC3339 to Unix nanosecond

Use the uint() function to convert an RFC3339 timestamp to a Unix nanosecond timestamp.

  1. uint(v: 2019-09-18T12:00:00.000000000Z)
  2. // Returns 1568808000000000000

Calculate the duration between two timestamps

Flux doesn’t support mathematical operations using time type values. To calculate the duration between two timestamps:

  1. Use the uint() function to convert each timestamp to a Unix nanosecond timestamp.
  2. Subtract one Unix nanosecond timestamp from the other.
  3. Use the duration() function to convert the result into a duration.
  1. time1 = uint(v: 2019-09-17T21:12:05Z)
  2. time2 = uint(v: 2019-09-18T22:16:35Z)
  3. duration(v: time2 - time1)
  4. // Returns 25h4m30s

Flux doesn’t support duration column types. To store a duration in a column, use the string() function to convert the duration to a string.

Retrieve the current time

Current UTC time

Use the now() function to return the current UTC time in RFC3339 format.

  1. now()

now() is cached at runtime, so all instances of now() in a Flux script return the same value.

Current system time

Import the system package and use the system.time() function to return the current system time of the host machine in RFC3339 format.

  1. import "system"
  2. system.time()

system.time() returns the time it is executed, so each instance of system.time() in a Flux script returns a unique value.

Normalize irregular timestamps

To normalize irregular timestamps, truncate all _time values to a specified unit with the truncateTimeColumn() function. This is useful in join() and pivot() operations where points should align by time, but timestamps vary slightly.

  1. data
  2. |> truncateTimeColumn(unit: 1m)

Input:

_time_value
2020-01-01T00:00:49Z2.0
2020-01-01T00:01:01Z1.9
2020-01-01T00:03:22Z1.8
2020-01-01T00:04:04Z1.9
2020-01-01T00:05:38Z2.1

Output:

_time_value
2020-01-01T00:00:00Z2.0
2020-01-01T00:01:00Z1.9
2020-01-01T00:03:00Z1.8
2020-01-01T00:04:00Z1.9
2020-01-01T00:05:00Z2.1

Use timestamps and durations together

Add a duration to a timestamp

date.add() adds a duration to a specified time and returns the resulting time.

  1. import "date"
  2. date.add(d: 6h, to: 2019-09-16T12:00:00Z)
  3. // Returns 2019-09-16T18:00:00.000000000Z

Subtract a duration from a timestamp

date.sub() subtracts a duration from a specified time and returns the resulting time.

  1. import "date"
  2. date.sub(d: 6h, from: 2019-09-16T12:00:00Z)
  3. // Returns 2019-09-16T06:00:00.000000000Z

Shift a timestamp forward or backward

The timeShift() function adds the specified duration of time to each value in time columns (_start, _stop, _time).

Shift forward in time:

  1. from(bucket: "example-bucket")
  2. |> range(start: -5m)
  3. |> timeShift(duration: 12h)

Shift backward in time:

  1. from(bucket: "example-bucket")
  2. |> range(start: -5m)
  3. |> timeShift(duration: -12h)