Removing point values with millisecs >0 in timestamp?

[Phillip Weeks]

This datasource is result of a persistent tcp connection which was synced more than once from the source. The original data source polls a water meter and the interval was quantized. The original data has never changed. I am wondering how it is possible to have a both a lower value and a greater timestamp if all data for this point emminates from the same ES persistent tcp connection? Might someone propose a explanation as to how this might occur?

Which leads me to my second question "Might anyone also know an easy method to selectively zap out all point values in a history which have timestamp > 0 millisecs?

[phildunlap]

Hi Phillip,

Hmm. I doubt that's what the original data set looks like. The ones at .000 are the real values?

Pretty small set of ways that can happen. Either

1. You are publishing two points to the same receiver with the same XID and so the values are being stored together.
2. You are publishing / saving real time data, while you are logging interval data. The interval logging maybe got off by 5ms due to some unknown delay while calculating its first delay. Then your receiver saved all the real time data and the publisher saved values at the interval log fire time. When a history sync happened, you had the minimum window size of some size, and the mismatch in some period got lots of values sent over with .005 because that's when the interval log task saved them.

We could have more information, like, what're the timestamps of the origin point? What is the publish event for the publisher, Update? What sort of logging is done on the source point?

I would sooner resync the series than delete every other value. You could purge the point, and then use a second publisher to perform the resync so that you didn't mess with the sync times for the real publisher.