InfluxDB' request Select where time > timestamp

Question

I have a java application that send measurements to an influxDB database.

I am adding points within my influxdb database. The points that I add each time that my program run have the current timestamp.

This is how I am adding my points (measurement building) :

    BatchPoints batchPoints;

    Date date = new Date();

    String beginofday = Constant.simpledateFormat.format(date);

    date = Constant.simpledateFormat.parse(beginofday);

    long timestamp = date.getTime();

    //adding points
    for buildings ... do
    Point point = Point.measurement("building").tag(tagsToAdd).fields(fieldsToAdd)
                        .time(timestamp, TimeUnit.NANOSECONDS).build();

    batchPoints.point(point);

My probem is that when I am requesting my database with such a request:

select count(tag) from building where time > (my timestamp)

I noticed that previous timestamp results are also counted, even that I am doing time>timestamp. And when I do a > instead of >= it counts only last ones. I noticed also that for the just previous timestamp, for example if I have timestamp like this 1540300800000 ns, influxdb add a 6 at the begining and it becomes 61540300800000 ms.

I really don't understand what is happening.

Any ideas?

Answer 1

java.time

    ZoneId zone = ZoneId.systemDefault();
    Instant beginofday = LocalDate.now(zone).atStartOfDay(zone).toInstant();

    long timestamp = beginofday.toEpochMilli();

    //adding points
    for buildings ... do
    Point point = Point.measurement("building").tag(tagsToAdd).fields(fieldsToAdd)
                        .time(timestamp, TimeUnit.MILLISECONDS).build();

I am using java.time, the modern Java date and time API.

What went wrong in your code

While I have not reproduced and tested, I believe that you confused milliseconds, 10^-3 seconds, and nanoseconds, 10^-9 seconds. date.getTime() gives you milliseconds since the epoch, but you are passing the number on in time(timestamp, TimeUnit.NANOSECONDS). If I take the start of the day today (Feb 21) in my time zone (CET) and use the milliseconds since the epoch as nanoseconds, I get 1970-01-01T00:25:50.703600Z. So I assume you got everything after some point around that time.

Other points:

• The classes Date and SimpleDateFormat that you used are poorly designed and long outdated, so I suggest you avoid them. The modern java.time is so much nicer to work with.
• Formatting your date-time into a date string and parsing it back is the detour for finding the start of the day.

Link

Oracle tutorial: Date Time explaining how to use java.time.