Cumulus Support

Hi

I have been using Cumulus with my Watson W-8681 for a few months now and find it great.
I upload to my website with no problems http://www.thatchmasters.ie/weather/

A couple of days ago I noticed a record temperature of 999, which was a tad too hot for the weather that was in it. Of course I had to mess with the files to try and put it right (oops), resulting in wild high rainfall (even for Donegal!) as well as temp records.

Is there anyway I can rectify the situation or must I start records from scratch?
What could have caused such a spurious reading - I hadn't even been near the whiskey that night!
Probably a corrupt file?

Thanks

Are you sure it was a max of 999 and not a max of -999, though that might sound even stranger? It's the default value for daily max. If normally immediately gets overwritten by the first real reading. If somehow the 'end of day' processing happens before a real value has been set, the -999 will get put into dayfile.txt for the daily max.

If you zip up your Diags files I may be able to see what happened (depending on how many times you've started Cumulus since then). Your PC clock doesn't set itself at midnight, by any chance (reminding me of another recent thread)?

One way you might be able to recover, again depending on how many times you've started Cumulus since the problem, is to use the most recent backup files from before the problem occurred. Cumulus will then re-download all the data since then from the station. Otherwise it will be a case of manually editing the log files.

Thanks Steve,

Yes maybe it was -999. I thought the sensor had maxed out , as last year I had a number of spurious rainfall readings in Easyweather.
Please find diags.zip attached.
It was more than likely my fault - maybe I asked the app to upload before vista was finished loading.
The rainfall data of 698.7 mm/hr might be the result of my messing too, either that or a spider has taken up residence in the sensor!

I just tried using the backup from before the trouble. The data were downloaded and all looked good until the max reappeared! See attachment screenshot.
It looks like the spurious reading is inside the Watson.
I checked clock update, it is set to sync at 5 p.m.
Thanks anyway

John, you have a corrupt dayfile.txt:

2009-09-04 17:34:44 : Error reading Dayfile: List index out of bounds (0)

It's probably a blank line or something like that - I assume this is one of the files you were editing?. This may well account for the strange values you're seeing. You will need to correct this - if the problem isn't obvious, upload it it and I'll have a look.

Steve

Yes Steve

The last lines in dayfile are all the same:
30/08/09,20.6,135,10:18,10.2,00:18,17.8,16:18,998.1,23:18,1010.5,00:18,8.4,21:18,7.2,14.3,104.0
31/08/09,12.1,360,08:49,11.5,03:49,18.5,18:17,989.6,22:47,998.1,00:18,3.6,01:25,12.6,13.8,57.3
01/09/09,15.2,270,12:47,10.9,01:47,16.7,13:17,988.6,05:17,991.3,13:17,0.6,00:47,0.3,12.3,50.7
02/09/09,9.8,270,00:17,999.0,00:17,-500.0,00:17,9999.0,00:17,991.3,00:17,0.0,00:17,0.3,16.7,0.0
03/09/09,9.8,270,00:17,999.0,00:17,-500.0,00:17,9999.0,00:17,991.3,00:17,0.0,00:17,0.3,16.7,0.0
04/09/09,9.8,270,00:17,999.0,00:17,-500.0,00:17,9999.0,00:17,991.3,00:17,0.0,00:17,0.3,16.7,0.0
04/09/09,9.8,270,00:17,999.0,00:17,-500.0,00:17,9999.0,00:17,991.3,00:17,0.0,00:17,0.3,16.7,0.0

Should I delete them or what?
Sorry to be trouble!

Yes, you may as well, as they're mostly garbage. Once you fix the problem with the file you can go back and retrieve the data again. It's easier to do it if you still have a back up, but you can also do it by manually editing the files back to a particular date. Did you find the problem with the file?

Hi Steve
I cant see any problem with dayfile.txt except for the last lines having the same values.
This morning I find that the lines have been duplicated (Not me mister, honest), so have attached same as a zip file.
Sorry to take up your time like this.
Regards
John

Yes, whatever the corruption was seems to have gone. So I can't explain why you're getting all these broken entries.

Could you please zip up your diags again and I'll have a look.

OK Steve, here they are

Your station rain counter has apparently gone haywire:

2009-09-06 11:32:21 : Writing todayfile, LastUpdateTime = 01/09/2009 13:18:21 raindaystart = 12650.7001953125
2009-09-06 11:32:21 : === Day reset complete
2009-09-06 11:32:21 : EWUSB: Ignoring bad data: raintot = 44499

Cumulus thinks the counter should be around 12650, but it now seems to be over 44000. It's not coping very well with this, and is discarding all of the data as invalid - this is a flaw in the bad data detection algorithm. It's basically got stuck, insisting that the rain counter should be much less than it is.

I'm not sure what the best way out of this is; it depends on how badly you want the missing entries in the log to be corrected. Basically, you need to edit the Start= value in today.ini to match the rain counter. If you just want to get going again, set it to something close to the current value of 44530 (44525 appears to be today's start of day value, so you could use that). If you want to go back and get Cumulus to correct the old data automatically, we'll have to try to work out what value the rain counter needs to be set to. But presumably the rain counter in the logger has this huge jump in it at some point, so we will have to try to find that and allow for it.

I reset the today.ini value as you suggested, Steve, and normality has returned.
I must check the rain unit for spiders.