Archive records read do not update graphs

[oh6hfx]

Sometimes my PC crashes or I need to restart it and I forget to start Cumulus.

Such a day was also yesterday, Cumulus was off around 10 hours when I noticed there is no icon in the
task bar.

Started Cumulus and waited a while, 595 archive records read (from WeatherlinkIP datalogger). Days records were updated and all seems ok again.

However I have noticed also earlier that the graphs are not updated - for example temperature graph is a straight line from the last "real time" measurement to the first
when Cumulus was started again. Should it be like that?

Johannes

[steve]

No, it should not. You may have a corruption in a data log, see: http://wiki.sandaysoft.com/a/FAQ#When_I ... rlier_data

But what you describe doesn't quite sound like that problem. If Cumulus was still running but unable to retrieve data, then it will continue to log the last data it received, so you will get flat line graphs. When you restart it, it wouldn't retrieve all of the missing data without manual intervention (restarting from one of the Cumulus backups from before the problem, for example). I'm afraid this is just the way Cumulus works, and at some point I may try to get it to behave better when it can't get data. There are some options available to work around this for now, e.g. you can get it to restart when data input stops (see the station settings). If it can't read data when it restarts, it won't log any data.

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

[oh6hfx]

Well when Cumulus (or my PC) crashed, it was still retrieving data ok from the Davis IP data logger. When I started PC again I forgot to start Cumulus, so all data was saved
in the datalogger for 10 hours. And when Cumulus was started again, it was downloading all the data ok from the logger, I could see the values changing in the Cumulus panel and all that days records were updated as well. Just the graphs were not ok.

Attached the files you asked..

Johannes

[steve]

It seems to have logged all of the data, there are no gaps, and no problems that I can see, so I can't say why you had the problem with the graphs. Are they OK now?

[oh6hfx]

Thanks Steve,

Now when I look graphs backwards again (select-a-graph) it seems to work. However that day when the crash happened (25 or 26th dec), that days actual graphs were straight line for the time when Cumulus was off. Paid not so much attention to that, maybe later (1h or so) the graphs were update ok then?

Johannes

[oh6hfx]

Problems with reading history data from the logger again.

Two days with occasional thunderstorms here and I have to take down my PC at home while WX station in remote location still stays on (hopefully) logging all the data
in the logger.

However, now when I have started Cumulus again after thunderstorm, seems that it does not read any data from the logger.

Today I updated to 1.9.4 and Cumulus was down 9 minutes form backup and install. Restarting, initializing, loading data, and result in the monthly log was this:

16.07.14;12:28;18,1;91;16,6;3,6;6,7;212;2,8;6,6;1010,91;128,8;28,2;50;4,5;16,8;18,1;0,0;0;0,00;0,00;17,8;697;0,0;196;0,0
16.07.14;12:29;18,2;91;16,7;3,6;6,7;210;2,3;6,6;1010,98;128,8;28,2;50;3,6;17,8;18,2;0,0;0;0,00;0,00;17,9;698;0,0;201;0,0
16.07.14;12:30;18,3;91;16,8;4,0;6,7;207;2,0;6,6;1010,98;128,8;28,2;50;4,5;17,1;18,3;0,0;0;0,00;0,00;17,8;699;0,0;209;0,0
.
.
16.07.14;12:39;18,6;91;17,1;3,1;4,5;206;1,0;6,6;1011,11;128,8;28,2;50;3,1;18,6;18,6;0,0;0;0,00;0,00;18,9;704;0,0;219;0,0;6,6
16.07.14;12:40;18,6;91;17,1;3,1;4,0;203;0,0;6,6;1011,08;128,8;28,2;50;2,2;18,6;18,6;0,0;0;0,00;0,00;18,9;704;0,0;219;0,0;6,6

Davis logger is uploading data also to Davis network (has been all the time). Could it be that this flushes the logger buffer?

Kind regards,

Johannes

[steve]

Davis logger is uploading data also to Davis network (has been all the time). Could it be that this flushes the logger buffer?

No, but what might have happened is that the logger was uploading to Davis at the time when Cumulus tried to download the archive data, so Cumulus was unable to connect to do that. I need to look at a 'retry' mechanism to cater for the way the IP logger works, but that will probably have to wait for the new version of Cumulus.

If you zip the diags folder up and attach it, I'll be able to see if that's what happened.

[oh6hfx]

Hi Steve,

Thanks, here is the diag zip for you:

https://www.dropbox.com/s/6x3d02to0qg0p26/cumulus.zip

Johannes

[steve]

I can't reconcile what I'm seeing in the diags files with what you say happened.

The .008 log file shows you closing Cumulus down at 12:30

16.7.2014 12:30:48.796 : Cumulus closing

Then the current file - .009 - shows you starting the same version of Cumulus up again at 18:04:

16.7.2014 18:04:11.468 : Cumulus 1.9.3 Build 1059 startup

16.7.2014 18:04:12.078 : Read today file: Date = 16.7.2014, LastUpdateTime = 16.7.2014 16:29:00, Month = 7

It had apparently been running earlier, as the last update time shows 16:29. But there is no diags file for that run. When Cumulus asked the logger for the entries it needed from 16:29 onwards, the logger started sending its entire contents:

16.7.2014 18:08:52.984 : Logger entry zero: Y = 2014, M = 7, D = 14, H = 19, M = 29
16.7.2014 18:08:52.984 : VP2: Earliest timestamp 14.7.2014 19:29:00
16.7.2014 18:08:52.984 : VP2: Earliest timestamp is earlier than requested
16.7.2014 18:08:52.984 : VP2: Find first entry after 16.7.2014 16:29:00
16.7.2014 18:08:52.984 : VP2: Checking logger entry 0 Y = 2014, M = 7, D = 14 H = 19, M = 29

At the end of the log, it was still downloading the logger contents, so I can't say what happened when it finally got to today's data

16.7.2014 18:08:53.031 : VP2: New earliest timestamp 15.7.2014 22:25:00
16.7.2014 18:08:53.031 : VP2: Checking logger entry 1617 Y = 2014, M = 7, D = 15 H = 22, M = 26

The logger will usually dump its entire contents if it's asked for an entry with a timestamp that it doesn't have exactly (rather than sending the data since that time, which would be sensible). So perhaps your console clock is wrong? The latest version does actually read the console clock and log it to the diags file, but you're still using 1.9.3.

[oh6hfx]

Thanks for your analysis Steve!

I had to revert back to 1.9.3 because 1.9.4 was producing .htm files of 0kt. Had no time now to start look what would cause that problem...
Lucky I had a backup from 1.9.3, just copied changed data files to that folder.

You are right, I should check the console clock. Actually it is not a console, it's Weather Envoy. Anyway, I will have a look...

Kind regards,

Johannes

[steve]

I had to revert back to 1.9.3 because 1.9.4 was producing .htm files of 0kt. Had no time now to start look what would cause that problem...
Lucky I had a backup from 1.9.3, just copied changed data files to that folder.

You need to read all of the release notes for 1.9.4, particularly the part about UTF8.

[sfws]

because 1.9.4 was producing .htm files of 0kt.

You need to read all of the release notes for 1.9.4, particularly the part about UTF-8.

You might find this FAQ http://wiki.sandaysoft.com/a/FAQ#My_web_pages_are_empty helps you to visualise what Steve describes in his release notes about UTF-8.
If you are using the standard templates, then 1.9.4 includes new versions of templates written to use the UTF-8 encoding.
If you have customised those templates, then the above referenced FAQ includes a link to further information about customised templates.

[oh6hfx]

Thanks Steve / sfws,

I'll have a look those UTF-8 things and try 1.9.4 again.

I have now looked WLIP web interface and all looks ok. Tough there is no setting for the clock. Don't know where it can be seen.

Also couple of weeks ago I changed batteries in Weather Envoy. Don't know if that could reset the date.

Kind regards,

Johannes

[steve]

Cumulus 1.9.4 will read the clock and log it to the diags file when it starts up.

[oh6hfx]

Managed to sort out 0kt htm problems and 1.9.4 running well.

I had to shut down PC during thunderstorm for 1,5 hours so good to test if it loads logger data now.

Seems not. In diag I can see that Cumulus tries to find a point in logger data where to start loading wx data?

18.7.2014 18:51:11.718 : VP2: Checking logger entry 2558 Y = 2014, M = 7, D = 18 H = 14, M = 49
18.7.2014 18:51:11.718 : VP2: New earliest timestamp 18.7.2014 14:49:00
18.7.2014 18:51:11.718 : VP2: No history records to process

18.7.2014 18:46:37.828 : VP2: Getting station clock
18.7.2014 18:46:39.093 : VP2: GetStationTime_V res = 0
18.7.2014 18:46:39.093 : y=2014 m=7 d=18 h=14 m=50

VP2 clock seems to be ~4h earlier than PC clock?

https://www.dropbox.com/s/wwu18t097s1z854/cumulus2.zip

Johannes