AnsweredAssumed Answered

Snapshot values not agreeing with archived values

Question asked by Ben_G on May 21, 2019
Latest reply on May 22, 2019 by Asle Frantzen

I've noticed an issue with snapshot values not lining up with archived values in PI.  Below is a screenshot of using the Compressed Data feature on Datalink with two tags. On the right, I have a tag from a plant that keeps track of which cathode is in a particular location.  This tag is a Float32, with Compression=1 and Exception=0.1, Step=1, Scan Class is 10 seconds, and it should always be an integer value.  In the Datalink screenshot, we can see that the cathode ID changes from 4 to 3 on 5/17/19 at 17:08.

 

On the left, I have a tag I created to store the snapshot values of the other tag.  The analysis that does this is shown below (the attribute 'CZ2_B2_CATNOW' references the tag 'CZ2_SB2_CAT_NUM'), and Debug_Tag has Compression turned off and Exception=0 so that it captures every value.  It is also Float32, and has Step=1.  In the Datalink screenshot, we can see that this tag changes from 4 to 3 on 5/17/19 at 16:02.

So, why do these tags change values at different times?  It seems to me that if the snapshot values on the left are correct, and compression is being applied as I understand it, the archived value should show the change happening at 16:02, and should record values immediately before and after the change.  I can't for the life of me figure out why these tags would differ like this, and this behavior is messing up the rest of my analysis.  Can anyone offer an explanation?

 

Furthermore, I notice that the archive values on the right often have a time interval of ~7 hours 50min.  Since the value should not be changing at all before it goes down to 3, and the Max. Time for compression is 8 hours, this also confuses me.  Does this have to do with the Max. Time for exception, which is 10 min?

 

In case it's relevant, the two tags shown here are from different PI servers in different time zones.  The original data comes from a plant in Oregon, and the analysis is being done on the server at our R&D center in Pennsylvania.

Outcomes