Data input error

[GraemeT]

Steve,
I've started getting this error each morning:

7/01/2012 2:03:00 AM : Data input appears to have stopped - check your station and connections
7/01/2012 3:03:00 AM : Data input appears to have stopped - check your station and connections
7/01/2012 4:03:00 AM : Data input appears to have stopped - check your station and connections
7/01/2012 5:03:00 AM : Data input appears to have stopped - check your station and connections
7/01/2012 6:03:00 AM : Data input appears to have stopped - check your station and connections

The error occurs 5 times only and always at the same time.

I haven't found any external causes, and am looking for suggestions as to what's wrong.


Any ideas?

[steve]

It can take over a minute to get data from a La Crosse station, particularly when it's busy trying to set its clock. This breaks the once-a-minute check in Cumulus. You can turn off the checking by editing cumulus.ini, and in the [Station] section add

ReportDataStoppedErrors=0

I should really turn it off automatically for La Crosse stations.

[GraemeT]

Thanks Steve,
That makes sense, since there doesn't seem to be any actual loss of data as a result.

What puzzles me is that the error only happens the 5 times and always at the same hour/minute...

Cheers.

[AllyCat]

Hi Graeme,

It's quite common for Radio Controlled Clocks to check approximately each hour throughout the night (when radio interference should be least) until they "find" a time signal. It's probably failing because I don't believe you have those signals in Australia. Can you turn the RCC function off?

Cheers, Alan.

[MikeLempriere]

I'm getting this error (Data input appears to have stopped) irregularly, but frequently, e.g. at least every few days, at most often twice in a day. As far as I can tell, I'm not losing data, thus I've learned to simply ignore this error.

Just to make sure -- this is Cumulus saying that it's not getting data from the base station, as opposed to the base station reporting communication problems with the weather station itself, correct?

If it's the latter, the weather station and base station are as close as they can be due to geographic constraints, so I'm just going to keep getting this occasionally. It's a new weather station (set up Mar-2012) and has shown this error as long as I remember, possibly since the beginning. However, as data seems fine, I don't think I have an equipment problem.

If it's the former, I have a theory as to why it may be happening. I'm using a very, very, very old IBM ThinkPad as the Cumulus computer. The computer is running Win XP, it is kept up to date on patches. It takes as much as 8 minutes to go from powerup to running Cumulus. I'm guessing that as opposed to a communication problem, what I'm really seeing is a short-on-horsepower problem. I'm guessing that Cumulus is getting swapped out while some other background process(es) run, and by the time it's back in, a notable amount of time has past. Some day if I can afford a newer computer, I'll try swapping this out with a faster one to see if the problem goes away, but that's not possible right now.

From the forum here, I see that I can disable the error msg, I will do that right away -- thank you for having made this configurable.

What I'd like to suggest is that the interval be made configurable as opposed to the blunt "disable". If I could change it to only alert on data failure for a longer duration, it would still notify me if there was a real problem (complete failure), but I could choose to "not sweat the small stuff". I'd suggest adding a .ini setting that takes a number of seconds of no data before warning.

Thanks for your consideration.

[steve]

Just to make sure -- this is Cumulus saying that it's not getting data from the base station, as opposed to the base station reporting communication problems with the weather station itself, correct?

That's right; it means Cumulus has been unable to read data from the station for over a minute. I don't think that a lack of PC power should be causing that, nor do I think that the period should be configurable to be longer - not for Davis stations, anyway. I think you probably are losing data, given that it should normally update every 2 to 3 seconds or so. But you would probably only notice this for peak wind gusts. If you zip up the diags folder and attach it, I'll see if I can see anything amiss.

[MikeLempriere]

Diag files as requested -- thank you for looking into it!

[steve]

I can't see what the cause is, but what is perhaps interesting is that certain times keep re-appearing, e.g 18:46:

6/24/2012 18:46:50.999 : *** Data input appears to have stopped
6/25/2012 18:46:04.078 : *** Data input appears to have stopped
6/26/2012 18:46:31.967 : *** Data input appears to have stopped

So I think the cause must be that your computer does something at particular times which stops Cumulus reading data (as you did in fact suggest). Or perhaps one of web site updates takes a long time to complete - do you upload to any of the 'third party' web sites like Weather Underground?

[MikeLempriere]

I recently uninstalled AVG antivirus, I'm thinking that may have been using up all the very limited processing time. I haven't seen the error since, but it hasn't been long enough for me to declare it as fixed...

Yes, I'm also pushing several additional files via ftp to my server, plus WUnderground.

[Super-T]

Latest AVG is very resource hungry especially on XP.
Try using Microsoft Essentials. Works for me.

[GraemeT]

I agree with you Terry.
All of the internet security suites such as AVG, McAfee and Nortons seem to swallow up resources, most noticeable on older machines.
Microsoft Security Essentials is about as lean (and cheap) as it gets and seems to do a good job of keeping out the gremlins.

[MikeLempriere]

Having uninstalled AVG seems to have stopped the problem. So my guess was probably good, the computer was simply bogging so badly, it got behind on data collection.

I do wish to point out though, I still use AVG on other computers, I believe it to be an excellent product, certainly on par with the "big boys".

In this case, this computer is within a firewall, and is dedicated to Cumulus, no email or web browsing, thus unlikely to get a virus unless it were to get it from another in-firewall computer -- all of which are running AVG or Norton.

Also, if the thing were to get a virus it would probably start to bog down again, and I'd starting seeing the data errors again!

Thanks all for your input/feedback.