Cumulus Support

All,

In need of a little help. Since upgrading to Cumulus 1.9.1 back in April I've been having problems with Cumulus closing/crashing. I get a message saying "Cumulus has experienced a problem and has had to close" (or something close to that). It then needs to be restarted. There seems no correlation between up time and a crash, can occur within a minute or after several hours. Any ideas?

Another issue is that on occasion Cumulus will download data and then on the very last line of data downloaded it'll come up with an extreme rainfall value. That then populates all my all time records, today's rain etc, etc. I have tried correcting this, both within the Cumulus text files and via the edit records option within Cumulus when it is running. No matter what I delete/correct the bad numbers re-appear next time I open the program.

Any thoughts on the above issues? I've pulled the upload to my website for the moment as I don't want bad data there. WeatherLink is unaffected by these issues - runs throughout and data is correct.

Edit: Running windows 7 and virtual VP

When Cumulus crashes, a bugreport.txt file is normally produced, could you upload it, please? If one isn't produced, then the crash is some sort of Windows issue.

Regarding the rainfall, it sounds like Cumulus is getting invalid archive entries back from Virtual VP, which I found did happen when I was testing with it. Virtual VP doesn't seem to like the Davis DLL which Cumulus uses. If you zip up the diags folder and attach it, I'll have a look.

steve wrote:When Cumulus crashes, a bugreport.txt file is normally produced, could you upload it, please? If one isn't produced, then the crash is some sort of Windows issue.

Regarding the rainfall, it sounds like Cumulus is getting invalid archive entries back from Virtual VP, which I found did happen when I was testing with it. Virtual VP doesn't seem to like the Davis DLL which Cumulus uses. If you zip up the diags folder and attach it, I'll have a look.

Hi Steve,

Sorry not got back sooner, been at work today. There is no bug report, for any of these shut downs, which makes it sound as if it is a Windows problem. It is odd though as this started as soon as I installed 1.91. TBH the frequent closing is an irritation but I can always restart - the odd v. high rainfall totals is more of a pain as it populates all my records and I have not managed to find a way to correct the data. If I can do that then it isn't a problem any longer. I sort of hope, although I do not know how realistically, that may be both these issues will go away on a later upgrade (maybe 1.9.2).

Zipped diags folder is attached. Thanks for your help.

I don't even know where to begin with all of the problems in the diags files!

Starting at the first file, cumulus.001, at 10/07/2011 23:34:12:

Cumulus thinks that when it last closed down, your annual rain counter was only 0.254. This is presumably because the problems had already started prior to that file.

10/07/2011 23:34:13.237 : Todayfileraindaystart = 0.254000008106232

The file suddenly terminates:

10/07/2011 23:34:41.754 : VP2: End of history download
10/07/2011 23:34:41.754 : PleaseWaitForm.Close executed
10/07/2011 23:34:41.895 : Wunderground archive entries = 4
10/07/2011 23:34:41.895 : Uploading Wunderground archive #1

I can't tell whether you killed it, or whether this is one of your crashes. It was about to update WU with an entry from the logger. Perhaps this is the cause of your crashes - some sort of problem accessing the internet at this point causing a crash in the Windows system code. You could try turning of WU catchup.

In cumulus.002, somehow the date in today.ini had been set back three days:

11/07/2011 19:43:43.099 : Read today file: Date = 08/07/2011, LastUpdateTime = 08/07/2011 14:16:00

I assume that you did this? There's no way that Cumulus could have invented that timestamp. You can't rewind more than about 1.75 days, because you have your station logger set to 1-minute intervals. But in any case, there was some kind of problem communicating with Virtual Vp or the station (this is one of the problems I saw when I was testing with VVP).

11/07/2011 19:43:51.461 : VP2: InitStation_V, res = -32701

In cumulus.003, Cumulus still thinks it needs to get three days worth of data:

11/07/2011 19:46:09.695 : Read today file: Date = 08/07/2011, LastUpdateTime = 08/07/2011 14:16:00

It can only get 1.75 days worth:

11/07/2011 19:46:24.546 : VP2: Earliest timestamp 10/07/2011 01:06:00

So it processes that data; but it's already had a lot of it, so this causes problems and the rain counter gets reset.

The file again terminates abruptly:

11/07/2011 19:51:00.791 : Creating WU URL #1384
11/07/2011 19:51:00.791 : http://rtupdate.wunderground.com/weathe ... 1&rtfreq=5
11/07/2011 19:51:00.978 : VP2: processing logger data entry 1384 for 11/07/2011 00:10:00
11/07/2011 19:51:00.978 : ih=46.0 oh=93.0 it=23.6 ot=9.0 wc=9.0 dp=7.4 pr=1020.221 wg=0.0 ws=0.0 wd=-32768 rt=0.00 rr=0.00
11/07/2011 19:51:00.978 : hi=9.0 thwi=9.0 thswi=-3.28E4 idp=5.8 ihi=22.9
11/07/2011 19:51:01.228 : Rain counter = 0

But note that it hadn't got as far as trying to send the data to WU in this case.

File cumulus.004 doesn't show any problems, but your rain counter had previously been reset, so the rain total for the day is large.

11/07/2011 23:28:00.803 : Writing today.ini, LastUpdateTime = 11/07/2011 23:28:00 raindaystart = 0 rain counter = 259.800018310547

In cumulus.005, the today.ini file had again been reset, so we start again with a reset rain counter:

11/07/2011 23:30:40.333 : Read today file: Date = 08/07/2011, LastUpdateTime = 08/07/2011 14:16:00

11/07/2011 23:30:55.247 : Rain counter = 0
11/07/2011 23:30:55.247 : 23:30:55 ****Rain reset? First chance: raindaystart = 256.199920654297, raintotal = 0

In cumulus.006, Cumulus hadn't been run for three days; this is a problem because the logger doesn't hold enough data to cover the gap, so the rain problem occurs again.

Moving on to the final file, cumulus.009 at 15/07/2011 19:02:29

It was looking OK here, once it had done the midnight rollover, it was back into sync with the rain:

15/07/2011 19:03:41.434 : Raintotal = 259.799987792969 Raindaystart = 0
15/07/2011 19:03:41.434 : Rainyesterday (calibrated) set to 259.799987792969
15/07/2011 19:03:41.434 : Raindaystart set to 259.799987792969
15/07/2011 19:03:41.605 : Writing today.ini, LastUpdateTime = 15/07/2011 raindaystart = 259.799987792969 rain counter = 259.799987792969

But then it had a problem communicating with Virtual VP:

15/07/2011 19:10:43.086 : VP2: LoadCurrentVantageData_V, error = -32701

And the WU updates weren't working:

15/07/2011 19:10:55.862 : An error has occurred during the Wunderground update: Request aborted on timeout

15/07/2011 19:11:32.395 : An error has occurred during the Wunderground update: can't resolve hostname to IP address

And then you closed it down before it had finished trying to update WU

15/07/2011 19:36:50.875 : Uploading Wunderground archive #88
15/07/2011 19:37:00.360 : Writing today.ini, LastUpdateTime = 15/07/2011 19:37:00 raindaystart = 259.799987792969 rain counter = 259.800018310547
15/07/2011 19:37:09.439 : MainForm can close
15/07/2011 19:37:09.439 : Showing splash screen
15/07/2011 19:37:10.312 : Exit FormCloseQuery
15/07/2011 19:37:10.312 : MainForm.FormClose
15/07/2011 19:37:10.312 : Terminating threads...
15/07/2011 19:37:21.326 : An error has occurred during the Wunderground update: Request aborted on timeout
15/07/2011 19:37:21.326 : Uploading Wunderground archive #89

I suggest you try turning off WU 'catchup' as it seems to cause problems on your system. Make sure you don't leave a gap of more than 1.75 days between runs of Cumulus, as your logger doesn't hold enough data for that. And don't attempt to 'rewind' (it looks like you were doing that, anyway) more than 1.75 days, and if you do rewind, use the backup files, don't just edit today.ini.

I'll try to make it handle the situation where the logger data doesn't cover the period since it was last running a little more gracefully (i.e. don't set the rain counter to zero) but it's always going to have problems making sense of the data if there are gaps that it can't fill in.

Hi Steve,

I frequently find that Cumulus doesn't close as it should. It will hang on the "Cumulus is closing this may take a few minutes" and will need to be forced to close. This might be what you are seeing in the diags and may be coincident with WU updates?

The re-winding more than 1.75 days is probably due to me trying to use back up files to restart the down load to rid the log of the excessively high rainfall value. I normally run Cumulus for at least 1/2 of everyday so gaps don't normally occur. On this occasion they did because I had Cumulus off line and was messing with different methods to remove erroneous rainfall value.

The editing and inadvertent screwing up of the .ini files, was a desperate, and failed, attempt to get rid of the rainfall issue.

I'll switch off WU catch up, can't recall if that was an option I activated in Cumulus or WU.

Is there a way of correcting the issues (esp. rain above) above? Cumulus is currently up to date - I've just removed its upload link at the moment which is why the website is out of date. The erroneous high rainfall value is high enough that it will remain for ever more.

Sorry about the state of the .ini files, I don't normally mess with the files - but the problem with the rain value prompted me to try and fix it and, as you can see, make a mess of them.

Steve,

For the minute I've taken the easy option of switching off Wunderground update completely in Cumulus - I'll take a look at disabling the catch up option when I am not working so much.

Steve Willington wrote:I frequently find that Cumulus doesn't close as it should. It will hang on the "Cumulus is closing this may take a few minutes" and will need to be forced to close. This might be what you are seeing in the diags and may be coincident with WU updates?

It hasn't hung, you're not giving it enough time to finish what it's doing. It takes a long time for the logger thread to finish writing everything away when it's had to download all of the entries from the log file. It takes even longer to send all of those entries to WU (version 1.9.2 has an indicator to show that it's still doing this).

Is there a way of correcting the issues (esp. rain above) above? Cumulus is currently up to date - I've just removed its upload link at the moment which is why the website is out of date. The erroneous high rainfall value is high enough that it will remain for ever more.

You can edit the log files to remove the erroneous rain (formats are in the help and the wiki). It's easier to use something like the toolbox (see 3rd party tools in the wiki) to do the editing. You can correct all-time highs using the built-in editor.

If you hit the problem again, it would be helpful if I could see the diags files before you start attempting to correct things. At the moment we have a 'chicken and egg' situation. All the problems I can see in the diags files appear to be a result of not running it often enough to prevent gaps in the data, and closing it down too soon after starting up with a large amount of data to download. I've changed the code today so that it no longer resets the rain counter when the logger data doesn't go back far enough.

OK. I'll have a look for that 3rd party toolbox software to sort my rain problems out - it'll be a few days before I get time.

As to the closing. What I normally see - say 3 times out of 4, is Cumulus posting that message "this may take a few minutes" and then close within seconds. On the times that it takes longer, it takes much longer. I can leave it 10 minutes or more and it will still be sat there with that message. It seems that if it doesn't close within a few seconds it does not close at all. Neither scenarios seem correct? I'll try leaving it for 1/2 hour next time to see if it does eventually close.

I always run Cumulus at least daily, as I do with WeatherLink, to ensure no missing data. As you say at present my mucking around blindly is preventing any proper diagnosis. I'll not mess with it next time and send you the diags as suggested.

Thanks for the help.

Steve Willington wrote:I can leave it 10 minutes or more and it will still be sat there with that message.

If it's downloaded 2500 logger entries and is trying to upload them to WU, it can take a very long time, particularly if it's struggling to contact WU.

For example, in the latest of the diags files that you uploaded, it started attempting to upload to WU the 1372 logger entries that it had just downloaded, at 19:10:25:

15/07/2011 19:10:25.442 : Uploading Wunderground archive #1

When you attempted to close it down 26 minutes later, it had only got to entry 88 (all had failed):

15/07/2011 19:36:50.875 : An error has occurred during the Wunderground update: can't resolve hostname to IP address
15/07/2011 19:36:50.875 : Uploading Wunderground archive #88

The problem is that failed uploads take much longer than successful uploads due to timeouts etc.

OK. Got it. I had no idea that uploading catchup data to WU took so long. Will be more careful in the future.

Is WU catchup something I can turn off? I did not see an option in Cumulus and I checked the WU site and I see no option there either. I vaguely remember switching it on, so will continue to look into it. For the minute WU uploads are turned off anyway.

Steve Willington wrote:OK. Got it. I had no idea that uploading catchup data to WU took so long. Will be more careful in the future.

Successful uploads are reasonably quick, typically less than a second per upload (but that still means half an hour or so for a full VP2 data logger). If there's a problem contacting WU, it takes a great deal longer. It should probably be more intelligent and give up. But that requires me to write yet more code.

Is WU catchup something I can turn off? I did not see an option in Cumulus

Only in 1.9.2.

I've downloaded and setup the 3rd party software "Toolbox". It's not clear to me which files I need to edit to correct my rainfall readings.

I've edited alltime.ini and also had a look through the July text file. The rain-counter contained therein drops to 0.0 on 11 July - but has the actual correct rainfall total for the year by the end of the file (today) 268.8mm. However Cumulus under rainfall displays a month total of 277.6mm and yearly total so far of 525.0mm.

So what to do? Editing the July file line by line would be a very lengthy process, 9 days worth of 1min entries. Any experts in using Toolbox out there?

Edit: BTW Steve - since turning off WU updates Cumulus has been stable, no closures or problems so far.

The most important file to correct is dayfile.txt, because the daily totals in there are used for all the other totals.

OK. Progress is finally being made. I've corrected the dayfile.txt, however there are 4 missing days in there, the 8th to 11th July inclusive. I do have 2 other dayfile.txts (I was backing up stuff whilst I tried to fix the rainfall total unsuccessfully) 1 of which has 8th and the other 9th included. Can I just cut and paste those 2 days into the current dayfile.txt?

The other 2 days saw no rain, but it would be good to recreate them - is there a way to reliably insert 2 days into the dayfile.txt? I can then manually file in the max/mins etc from WeatherLink via the Toolbox app.

Steve Willington wrote:1 of which has 8th and the other 9th included. Can I just cut and paste those 2 days into the current dayfile.txt?

Yes.

The other 2 days saw no rain, but it would be good to recreate them - is there a way to reliably insert 2 days into the dayfile.txt? I can then manually file in the max/mins etc from WeatherLink via the Toolbox app.

The easiest way to insert two new days would probably be to copy two existing ones and then edit using Toolbox.