Vinod said
Hi Geoff,Now that you have described how the channel is populated, I know what is going on.
When you clear the channel and have the devices continue to upload to it, if the device has uploaded data before you have imported the old data, the entries in the database are not monotonically increasing from the creation date of the entry. To take an example, lets say you had 2 entries posted via the API at 11:00 AM and 12:00 AM to a freshly cleared channel, these two would be entry (1) and entry (2) which have a date of today. Now, you imported your data from a CSV that had 200 rows. The entries # 3-#202 in the database would actually have been measured before entry (1) and (2). This exposes a bug in the graphing algorithm when you do daily/monthly sums which makes the assumption that the data is monotonically increasing. It is basically the same out of order dataset which causes the "data export" in the private view to return partial data. Thank you for reporting this. We'll work on resolving this on our end.
There is a workaround you can do in the mean time - export your data from the "Import/Export" tab. Clear your channel and immediately re-import the data. Then allow your devices to continue to post data to the channel. The graphs should look correct now.
-Vinod
Thanks Vinod. I'll give it a shot to export the data, clear the channel and re-import it.
That leaves me with one question: is there a way I can address the timezone issue to have it do the daily sums from 0:00 Australia/Melbourne rather than starting the daily sums at 0:00 UTC?
Regards,
Geoff.