Max min graph not working correctly.

[Bolty]

I am having some problems with the graphing of max/min temperatures.
Both my website and the graph option within Cumulus display incorrectly. I have attached a jpg file showing the incorrect graphic.

I have checked the dayfile.txt file and have found the log contains temperatures for the peaks in the graph as 999 max and - 500 min. Therefore the graph is probably correct, just the data is incorrect.

I can manually edit out these errors, but I was wondering what would most likely have been the cause of these false values. I assume that the Lacrosse WS reports these and Cumulus just records them. In that case Cumulus is working perfectly well, and the fault lies with the WS. On the other hand, if Cumulus generates these values from it's own monthly log file, then I need to sort out the cause.

Has anybody got any ideas on what could be the cause, and has anybody else experienced this problem.

Many thanks to any ideas

[steve]

I have checked the dayfile.txt file and have found the log contains temperatures for the peaks in the graph as 999 max and - 500 min.

It's the other way around (999 min and -500 max). Those are the default start of day values. The idea is that as soon as the first real temperature value is read, those values are replaced by the first real values, so there's no special code to cope with not having a previous value to check against.

The way that those values could get logged in dayfile.txt would be where Cumulus did its end of day processing without having had any real data to replace those figures. The tricky bit is working out what sequence of events could cause that to happen. One situation which used to cause problems is when you start and/or stop Cumulus close to midnight. It should cope, of course, and I've plugged most of the holes in that area, I think.

What do the other values look like in those entries where these figures occur? Is there more than one entry for the same day (looks like there are two entries for 4th August)? If you attach the file here, I can have a look and maybe spot something. If you still have files in the diags folder which are from around the 9th August (when it last occurred) those would be useful too.

[Bolty]

Thanks very much Steve for your help!
Since writing the post, I have corrected the dayfile.txt manually, and unfortunately I did not keep the old dayfile. However I have copied the last few days of the dayfile from 10th August which does show the data that I removed.

You are quite correct in stating that there are duplicate entries. This is probably the cause of the problem I suspect. You may recall that I emailed you personally with similar problems a few months ago! There were then duplicate entries which I put down to running Cumulus as a service on WHS. At that time, without realizing that it caused problems, I would concurrently try to run Cumulus as an application by remote desktop. I am almost certain that that was the cause of the duplications. Now I can understand from your description, that the max min being 999 and -500 come from no data entries for that day. You will see from the attachment below that the problem did occur on the 5th and again on the 9th.

I think that the lesson for me may well be that I must first stop the Cumulus service any time when I want to access Cumulus directly. I hope that this will avoid the duplication.

I know that you also run Cumulus as a service on WHS. I therefore wonder how you deal with the issue of being able to look at Cumulus info from "select a graph" whilst Cumulus is running as an invisible service on the WHS? I do not really know how to look at historic data from Cumulus when WHS is running (and continually logging) Cumulus as a service.

I wonder if it is possible to easily modify Cumulus to run a separate (and concurrent) READABLE ONLY instance of Cumulus either from within WHS or a remote desktop. This instance of Cumulus would not need (or be able) to read the serial port or current observations, merely use the data from the logs to present graphs etc. It would be a read only version of the application that could not modify the log files. I see this as being a similar approach to the way in which the web site works, except that with the web, you are unable to look at historic data.

These may be crazy concepts, but I present them merely as ideas. You may well already have the solution in hand with a different approach.

As always your energies and talents continue to amaze me!

Thanks very much!

[Bolty]

Woops, sorry! Forgot to append the part of the dayfile.

03/08/09,6.9,337,23:22,8.3,09:00,16.9,14:08,1018.4,04:36,1026.0,09:43,0.0,09:00,0.0,12.4,67.9
04/08/09,9.9,292,14:22,11.9,22:54,17.2,14:31,1018.0,14:47,1020.4,10:22,4.1,11:29,5.2,14.2,54.3
05/08/09,0.0,0,09:00,999.0,09:00,-500.0,09:00,9999.0,09:00,0.0,09:00,0.0,09:00,5.2,0.0,0.0
05/08/09,0.0,0,08:18,999.0,08:18,-500.0,08:18,9999.0,08:18,0.0,08:18,0.0,08:18,5.2,0.0,0.0
04/08/09,1.4,292,01:06,7.8,06:06,12.1,00:06,1018.6,03:06,1019.5,08:06,0.0,09:00,5.2,9.4,5.0
05/08/09,6.7,248,07:06,8.1,02:06,18.7,13:06,1010.7,06:06,1019.7,09:06,0.0,09:06,0.0,12.6,135.0
06/08/09,24.3,337,13:45,8.9,07:29,20.1,13:14,1005.4,16:22,1021.7,08:59,2.6,15:39,7.3,14.9,377.8
07/08/09,9.1,202,10:16,2.0,07:17,16.1,14:58,1021.6,09:01,1025.0,19:54,0.5,10:30,0.5,8.5,39.1
08/08/09,6.3,360,14:25,5.2,09:00,16.9,14:03,1018.9,15:10,1023.5,09:00,0.0,09:00,0.0,14.9,134.9
09/08/09,2.4,315,09:00,999.0,09:00,-500.0,09:00,9999.0,09:00,1019.0,09:00,0.0,09:00,0.0,15.8,167.9

[steve]

I know that you also run Cumulus as a service on WHS. I therefore wonder how you deal with the issue of being able to look at Cumulus info from "select a graph" whilst Cumulus is running as an invisible service on the WHS? I do not really know how to look at historic data from Cumulus when WHS is running (and continually logging) Cumulus as a service.

I just run it as a normal application, not a service. It's a pain when WHS reboots after an update (as it did earlier today), but it doesn't happen too often and I soon notice it.

I wonder if it is possible to easily modify Cumulus to run a separate (and concurrent) READABLE ONLY instance of Cumulus either from within WHS or a remote desktop. This instance of Cumulus would not need (or be able) to read the serial port or current observations, merely use the data from the logs to present graphs etc. It would be a read only version of the application that could not modify the log files.

Possible, yes, and hopefully I'll add something like this to Cumulus 2 at some point.

[steve]

04/08/09,9.9,292,14:22,11.9,22:54,17.2,14:31,1018.0,14:47,1020.4,10:22,4.1,11:29,5.2,14.2,54.3
05/08/09,0.0,0,09:00,999.0,09:00,-500.0,09:00,9999.0,09:00,0.0,09:00,0.0,09:00,5.2,0.0,0.0
05/08/09,0.0,0,08:18,999.0,08:18,-500.0,08:18,9999.0,08:18,0.0,08:18,0.0,08:18,5.2,0.0,0.0
04/08/09,1.4,292,01:06,7.8,06:06,12.1,00:06,1018.6,03:06,1019.5,08:06,0.0,09:00,5.2,9.4,5.0
05/08/09,6.7,248,07:06,8.1,02:06,18.7,13:06,1010.7,06:06,1019.7,09:06,0.0,09:06,0.0,12.6,135.0

That part is quite interesting! I can't think of any bug in Cumulus that could cause that sequence. I think it would have to be down to running two instances at once.

[Bolty]

Thanks a million Steve!

I agree with you that it is due to 2 instances running. It is pleasing that Cumulus is not the cause.

Running Cumulus as service is really very good, especially when updates come through, as Cumulus then restarts after the WHS reboots. (So long as I remember to stop the service if a run a separate instance)

If you thought it helpful for your web site, I would be happy to document the process that I used to get it operating as a service. I would write the document in a very clear step by step set of instructions if required!

[steve]

If you thought it helpful for your web site, I would be happy to document the process that I used to get it operating as a service. I would write the document in a very clear step by step set of instructions if required!

That would be great - in fact I think someone else has been asking how to do it in another thread. I might even give it a go myself!