Double entries in dayfile.txt?

[Gordon-Loomberah]

I was just looking at my average temperatures graph and noticed that 28/8 seemed to have 2 readings for the day- as there is a vertical line for that day. http://members.ozemail.com.au/~loombera ... trends.htm
Checking dayfile.txt I see there are 2 entries for the 28th. I have seen that once before, I think I was editing something at 9am, but I dont think that was the case on the 28th.
Any ideas on what the cause may be?
Is there information somewhere that shows the format of that file to save having to go through and work it out? I'd like to check that I am deleting the correct line and that the 2nd line doesnt have part of the info that should be in the line I keep.

thanks, Gordon

[beteljuice]

Is there information somewhere that shows the format of that file to save having to go through and work it out

That would be Help / Data Log Files or http://wiki.sandaysoft.com/index.php?title=Dayfile.txt

[Gordon-Loomberah]

Ahh thanks, I should have looked at the Wiki before asking in a rush when I had to go out

It seems the top line for the 28th was correct, the 2nd one was made up from readings after 9am on the next day. I wonder how that could happen

thanks, Gordon

[steve]

Do you use the midnight to midnight day, or 9am to 9am?

If you zip up the diags file and attach them, I might be able to work out what happened.

[Gordon-Loomberah]

I'm using 9am to 9am.

cumulus.zip

[steve]

It did the day rollover at 0900 and then again at 1000, very odd. Were you changing the setting which says use 1000 in summer? But there again, I don't suppose you've gone onto summer time yet anyway? It's as if it got the wrong answer back from Windows when it asked for the timezone settings.

[Gordon-Loomberah]

No I didnt change that, and dont plan to in October when we change to daylight savings either.

thanks, Gordon

[steve]

I see what's happened now:

2009-08-29 10:00:00 : Writing todayfile, LastUpdateTime = 29/08/2009 10:00:00 raindaystart = 2.58999991416931
2009-08-29 09:59:54 : === Day reset for 29/08/2009 09:59:54
2009-08-29 09:59:54 : Raintotal = 2.58999991416931 Raindaystart = 2.58999991416931
2009-08-29 09:59:55 : Rainyesterday (calibrated) set to 0
2009-08-29 09:59:55 : Raindaystart set to 2.58999991416931
2009-08-29 09:59:55 : Writing todayfile, LastUpdateTime = 29/08/2009 10:00:00 raindaystart = 2.58999991416931
2009-08-29 09:59:55 : === Day reset complete

Your PC clock has gone backwards. Every time the hour changes, Cumulus checks to see if it is time to do the day rollover. At 10:00:00 it started to do the check for that hour, but the clock then changed and the hour was 9 so it thought it was 09:00 and did the rollover. I should probably have some checks in for this situation. I'm guessing your system adjusts its clock at 10:00?

[beteljuice]

Every time the hour changes, Cumulus checks to see if it is time to ...

Might be worthy of an Important Mention somewhere ?

Close Cumulus when adjusting / changing PC clock !

[steve]

Every time the hour changes, Cumulus checks to see if it is time to ...

Might be worthy of an Important Mention somewhere ?

Close Cumulus when adjusting / changing PC clock !

If you're adjusting it manually, then it doesn't matter whether you stop Cumulus or not. If you move it back to the previous meteorological day, I think you might expect Cumulus to get a bit confused, and it will, but it doesn't matter whether it's stopped or not. The problem in this case is because the time was being adjusted automatically at exactly one hour after the end of the meteorological day.

The advice perhaps should be to make sure your PC isn't set to adjust the clock automatically within an hour of the end of your meteorological day.

[Gordon-Loomberah]

OK, that reminded me...I did a manual internet sync of it at 10:00:02am, and it went back to 9:59:54, I was watching the clock to see how much it lost in the week since the previous adjustment.

I wont be adjusting it manually before 10am from now on!

thanks, Gordon

[steve]

If you did it manually, that was quite some timing! You managed to get in after the 'hour changed' event was fired in Cumulus, but before it processed it.