0
votes

I am currently sending logs to Graphite that are updated with timestamps and a value every 30 minutes, however I am also keeping logs that calculate trends in this data for every 30 minute period over each day in the course of a week.

For example, in JSON:

{
            "stat_id": "1-0-32",
            "_t": "16:00",
            "value": 26.0,
            "value_count": 4.0,
            "day_type": "Weekday",
            "day": "Monday"
          }

would be the average value for 16:00 on a Monday for a specific metric. I would like to be able to graph this trend data on top of the raw data that is already being graphed. I've already used Graphite functions like averageSeries(timeStack(metric, "7d", x, y)) to try to recreate the trend data but am not confident it represents the trends as accurately and completely.

So I guess the question is can I do this by sending data to Graphite that only changes in value and where the timestamps are static and somehow represents hours in the week?

1
I suppose one thing I can do is to give this data arbitrary timestamps that represent some week and then use absolute time ranges to merge over my raw data... - dsapoz

1 Answers

0
votes

What I ended up doing is adding the trend data as if it were real-time data by applying the time stamps to the current day, making sure not to add future metrics.