Daylight savings end

[DaveNZ]

Hi Steve, I can't remember what my graphs looked like the last time Daylight savings ended, but is this normal behaviour?

The clock went back at 3:00am, and Cumulus seems to have graphed the data as

2:00 (old), 2:00 (new), 2:01 (old), 2:01 (new), etc...

Example: pressure graph

press_endDST.png

Zooming in on the thick area...

press_endDST_zoom.png

[steve]

The graph component handles the actual plotting, Cumulus just gives it the data points. Cumulus will have fed the data into charts in the correct order - 0200 to 0300, and then 0200 to 0300 again, but the graph is being 'clever' and plotting the points in strict chronological order according to the timestamps.

[tjaliwalpa]

This morning my graphs looked similar. I realised they are merely plots of the data. I shut down Cumulus, restored a backup from before midnight and changed the console clock to match my PC clock.

When I restarted Cumulus, all was smooth and lovely

I also had a weird drop in pressure that didn't make sense and using the restore of data is the usual way I correct that situation and get my pressures back to were they are meant to be.

[Super-T]

Ditto here
https://cumulus.hosiene.co.uk/viewtopic.php?f=20&t=4976

Perhaps it is better to do a shutdown/restart of Cumulus the night before Daylight Saving ends and then do a recovery the day after?

How many others had this problem last night?

[steve]

Cumulus has always been like this; I don't understand why suddenly this year there are all these queries about it. As I've just said in response to Terry's post, if you 'rewind' and force Cumulus to re-timestamp everything, you're actually putting incorrect timestamps on your data. The best way is to leave it running and understand that Cumulus timestamps using 'local' time, so you will have two sets of data for an hour with the same timestamps.

Even if it timestamped using UTC (or local time ignoring daylight savings), the graphs would still look odd for an hour - unless you were happy to see the graphs plotted ignoring DST all the time.

Steve

[DaveNZ]

I'm not complaining, just couldn't remember it being like that, after all it only happens once a year Which I can deal with...

[steve]

I'm not complaining, just couldn't remember it being like that, after all it only happens once a year Which I can deal with...

I did wonder if the current version of the graphs component draws the graphs slightly differently, but I think they've always looked like that.

[Gina]

There is always bound an anomaly in graphs around the days the clocks are changed whatever software you use - unless you stick to GMT/UTC or local time without DST (or BST in the UK). I log in GMT/UTC but graphs are shown in local/BST. It is not the fault of the software if graphs are strange when we mess with our clocks.

It might be possible to draw special graphs showing the overlap or the gap, but since this occurs just twice a year for a few hours, it's not really worth the bother. Programming time and effort is better spent on other things IMO.