Problems with Cumulus closing and high rainfall totals

[Steve Willington]

Hi Steve,

This is what I did when Cumulus didn't initialise properly.
Closed Cumulus.
Opened windows explorer and found the Cumulus backup folder.
I then copied ALL the files in the penultimate backup folder (20110730084526), assuming that the most recent (20110731190046) was a backup saved when closing Cumulus after it failed to initialise correctly, to the data folder (over writing the version in data folder).
Then restarted Cumulus and it is then that the spurious rainfall value of 270mm appeared. This must be when cumulus.ini got wound back? Should I not copy all files in those backup folders across to the data folder?

The time between Cumulus sessions was less than 24 hours (probably 18). If there is a gap it is because I went back to that penultimate backup folder. I assumed that the date/time of the folder was when the previous (successful) Cumulus session was started and that the data in it would extend to the time I closed Cumulus the previous evening. Perhaps this is not the case and the date/time is the date/time of the most recent data backed up? When does Cumulus create these backups? I thought when Cumulus is closed down?

What I may do is take the website off-line until 1.9.2 is available and see if that solves the problem. Sadly I've not got the time to spend correcting these things (wish I had). The philosophy is just to set it up and leave it doing its thing. The site is really for my own interest, made available for others interested in the weather, so I could just password the site to limit access to those aware that the odd data issue may occur. The occasional data error would not matter then and I could get around to fixing when I have time.

[steve]

The backups are taken when Cumulus starts, so if you have a problem with incorrect data during a particular run, you can start that run again by using the backup files in the folder with the timestamp of that run.

However, when Cumulus can't talk to the station, it doesn't change any data, so there is no need to use a backup to rewind. The backups are for when you have a problem with your data; you did not. You then used a backup from too far back such that the data in your logger did not cover the missing period, and that is what caused the problem with your data. Cumulus is always going to have problems recording correct figures when there is a large gap in the available data, particularly when the gap affects more than one day, so it spans midnight.

Version 1.9.2 is available any time you want to use it. But neither 1.9.1 nor 1.9.2 will be able to give you correct results if you rewind the data unnecessarily and rewind beyond the limits of your logger. Hopefully, now that you understand more about the backups, and when and how to use them, you won't have any more problems.

[Steve Willington]

Hi Steve,

I feel your frustration - for the record I am absolutely certain that the problems I am having are largely (solely) user related and not Cumulus. With a better understanding of the purpose of the backup files (I was assuming that Cumulus would start downloading data from the time of the last close down) I suspect that the rainfall issue, once solved this time, will go away.

The rainfall issue is doing its best to persist this time though. I have just fired up Cumulus again, in the hope that the 270mm would appear in yesterdays dayfile.txt, so I could edit it out with toolbox. It does not - it has appeared in today's data again - yesterdays records in Cumulus show no rainfall, which is correct. I'll try again tomorrow morning to see if I can then edit 01/08/11 (today's) data in dayfile.txt

[Steve Willington]

Well I wish I knew what was going on. I fired up Cumulus this morning and, as expected, the 270mm appeared in yesterdays total. So I have just spent an hour correcting the files via toolbox, including the minute rainfall count and rainfall so far (which is what took me so long).

However on firing up Cumulus the 270mm has re-appeared in the last data line in today's total???? Where is it coming from? Maybe I shouldn't have corrected the minute data (rainfall count and total rain so far)? I now have to wait until tomorrow to try again.

[steve]

If you zip up the diags folder and attach it, along with your August log file, I'll have a look.

[Steve Willington]

Steve,

I've re-corrected the data as best I can using toolbox. Attached is the diags folder and August log file - hopefully you'll be able to spot what I am missing.

[steve]

Something happened to the 'start of day rain counter' in today.ini between closing Cumulus down at 09:38

02/08/2011 09:38:00.831 : Writing today.ini, LastUpdateTime = 02/08/2011 09:38:00 raindaystart = 270.200012207031 rain counter = 270.200012207031

and starting it up again at 10:05

02/08/2011 10:05:38.347 : Read today file: Date = 02/08/2011, LastUpdateTime = 02/08/2011 09:38:00
02/08/2011 10:05:38.378 : Todayfileraindaystart = 0.200000002980232

It had the correct value when it closed (270.2) but it looks like you edited it to 0.2? This is why you now have a rain for today of 270 (270.2 minus 0.2), whereas before you edited the file it had today's total correct.

[Steve Willington]

Arrggghhh - I see. I made the mistake, of assuming that every 270mm was an erroneous one. My actual yearly total is also 270mm + the 0.2mm of yesterday, or 270.2mm. When I saw that 270.2mm number I changed it back to yesterdays total of 0.2mm by mistake.....

So, lets get this right this time (me not you) - what number(s) do I need to change back to get this running right on the next Cumulus start-up?

Thanks for spotting the error on my part - I now imagine that the erroneous rainfall figures I have been seeing occasionally is my yearly running total to that date???? Sorta shows that I don't follow my own AWS stats that closely

[steve]

Cumulus uses a 'total rain counter' supplied by the station to work out all its rain data. In the case of the VP2, it uses the annual total for this purpose. It stores the value of this counter in today.ini at midnight, so that during the day it can subtract that stored value from the current annual total, and hence determine today's total so far.

So you need to edit today.ini and set the Start= figure back to 270.2 before you next run Cumulus. Today's total should then be correct again.

[Steve Willington]

Success!

[Steve Willington]

Hi Steve,

I've recently upgraded to Cumulus 1.92 and turned Wunderground updates/catchup back on. Cumulus crashes again, but does so in a more orderly manner, offering a report and the choice to restart or close. Not sure if it will help but I've attached a bug report.

[steve]

Sorry, I've no idea what the cause of the crash is, it's crashing in Windows (ntdll.dll). Possibly a driver problem?