Cumulus Support

For some unexplained reason the clock on my PC has a mind of it's own & occasionally shows the year as being completely wrong. An example of this was yesterday evening when starting up it showed the date as 18th June 2015. I usually check this when starting but last night I seemed to have missed the incorrect year, yet the time was correct.

I then started WeatherLink & it down loaded the console as usual & all the information was correct & no data had been altered or corrupted & nothing showed up even when I checked the database. So at this point I was not aware of any date problem with the computer. I then closed WL.

Then I opened Cumulus & of course data was in a mess as it read the computer date & of course there was no data other that what had just been down loaded from the console. With some thought & the use of Notepad++ I was able to correct it all by editing the today.ini & the yesterday.ini files, so all it well now.

My reason for posting this observation is that it seems WL uses the console clock for all data collection & filing so is somewhat immune to PC clock variations, whereas Cumulus uses the PC clock for it's data collection & filing & is very prone to any problems in that area. While this problem was my fault I thought perhaps this observation may help when diagnosing problems with data & times between Cumulus & the console.

Fortunately I did not have Cumulus setting the console clock in this case. I do not run my computer 24 hours a day & I usually only check WL & Cumulus once or twice a day, so please don't use this to criticize me in the way I use my computer.

The difference between WL and Cumulus is that WL was written specifically for Davis stations, and only for Davis stations. The WL logs consist of data downloaded from the station's logger periodically, so they contain the timestamps from the logger.

Cumulus is not specific to Davis stations, and hence works differently. While it is running, and logging 'live' data, it does not use the station logger at all, it constructs its log entries from the live data that it has been sampling, and timestamps each log entry using the PC clock. When it starts up and needs to catch up using the station logger, it behaves in manner similar to WL. It constructs entries in its own tables from entries downloaded from the logger, and it timestamps those entries using the timestamp in each logger entry. But it uses the PC clock to determine which entries it needs to download from the station. It can't work any other way.

For some unexplained reason the clock on my PC has a mind of it's own & occasionally shows the year as being completely wrong. An example of this was yesterday evening when starting up it showed the date as 18th June 2015. I usually check this when starting but last night I seemed to have missed the incorrect year, yet the time was correct.

How old is your computer as usually this is a symptom of a BIOS battery on the motherboard not maintaining a charge properly and therefore the date is not maintained correctly? Very simple to change as long as you know how to open the case.

steve wrote:The difference between WL and Cumulus is that WL was written specifically for Davis stations, and only for Davis stations. The WL logs consist of data downloaded from the station's logger periodically, so they contain the timestamps from the logger.

Cumulus is not specific to Davis stations, and hence works differently. While it is running, and logging 'live' data, it does not use the station logger at all, it constructs its log entries from the live data that it has been sampling, and timestamps each log entry using the PC clock. When it starts up and needs to catch up using the station logger, it behaves in manner similar to WL. It constructs entries in its own tables from entries downloaded from the logger, and it timestamps those entries using the timestamp in each logger entry. But it uses the PC clock to determine which entries it needs to download from the station. It can't work any other way.

Thanks Steve. I fully realise that & the purpose of my post was to perhaps help explain some of the time variations that are often mentioned in posts in relation to temperature & other items concerning Davis stations. It was not meant as a criticism of Cumulus.
Bligh.

water01 wrote:

How old is your computer as usually this is a symptom of a BIOS battery on the motherboard not maintaining a charge properly and therefore the date is not maintained correctly? Very simple to change as long as you know how to open the case.

Thanks David. My computer is 5 years old & I have replaced the BIOS battery twice the last one being earlier this year, so that is not the problem. It has been something of a problem really ever since I have had it & of course only really revealed itself after the warranty had expired.
Bligh.

Touchtone wrote:It was not meant as a criticism of Cumulus.

Yes, understood. My reply wasn't meant as a defence to a perceived criticism, just an explanation.

Couple of things come to mind here.

1. Make sure the Windows Time service is running on your computer and set to automatic on startup.

2. Don't use time.windows.com (Windows default) as your NTP server as it is unreliable even in the states. Use one closer to Oz, like au.pool.ntp.org

Good luck,

--Kurt

N0BGS wrote:Couple of things come to mind here.

1. Make sure the Windows Time service is running on your computer and set to automatic on startup.

2. Don't use time.windows.com (Windows default) as your NTP server as it is unreliable even in the states. Use one closer to Oz, like au.pool.ntp.org

Good luck,

--Kurt

Thanks Kurt. I do have Windows Time running & I don't have have an NTP server as it is a stand alone computer.

Since I posted this subject I have found that the computer power supply has a leaking capacitor, so I will get a new PSU tomorrow & replace it. I am sure that is the cause & also the fan it very noisy, probably a dry bearing, so a new one is called for. Hopefully that will end this irritating problem.

Bligh.

steve wrote:The difference between WL and Cumulus is that WL was written specifically for Davis stations, and only for Davis stations. The WL logs consist of data downloaded from the station's logger periodically, so they contain the timestamps from the logger.

Cumulus is not specific to Davis stations, and hence works differently. While it is running, and logging 'live' data, it does not use the station logger at all, it constructs its log entries from the live data that it has been sampling, and timestamps each log entry using the PC clock. When it starts up and needs to catch up using the station logger, it behaves in manner similar to WL. It constructs entries in its own tables from entries downloaded from the logger, and it timestamps those entries using the timestamp in each logger entry. But it uses the PC clock to determine which entries it needs to download from the station. It can't work any other way.

Hallo Steve,

I do have also a question concerning the timestamp.
I understand Cumulus is not reading the date and time from the Davis logger, maybe it's possible to explain why not?

My Davis VP2 console is running on UTC time, am using Virtual VP, so also using Weatherlink.
In Weatherlink the records stored are in UTC time, my PC runs in Local Time, and I like also to have Cumulus stores the records in UTC time, if possible direct from the logger.
Or maybe there is a parameter in Cumulus, to make an offsettime to the PC time, or the timezone.

If possible I like to have my PC running on local time.
If it's not possible to configure Cumulus so the records are stored in UTC time, it's the only option.

Maybe you can tell me your creative thouhghts.

Thanks in advance.

Peter

pb2rdf wrote:I understand Cumulus is not reading the date and time from the Davis logger, maybe it's possible to explain why not?

It does use the date and time from the logger, when it's using entries from the logger, i.e to catch up when it starts up. Each logger entry has a timestamp in it, so it makes sense to use that. When it's recording live data it uses the PC clock. It wouldn't make sense to have to keep reading the console clock, it would lose data while it was doing that. The only way to make it log live data using UTC is to run the PC on UTC.

I am not sure what programming language is used for cumulus but I would have thought it would be relatively easy to have a check box which a allows a user the opt in for storing data in UTC by calling a function that converts the local pic time to UTC.

I might have it all wrong but I just thought I would chime in on the conversation.

Except that I am no longer doing enhancements to Cumulus 1. I am explaining the situation as it stands in Cumulus 1, as this enquiry is about Cumulus 1.

This might be an option in Cumulus MX at some point in the future. But it always makes me laugh when someone says "it should be easy...". I guess I'm just lazy and stupid, I should have added all of the hundreds of outstanding enhancement requests to Cumulus long ago

I did not mean to offend and I am sorry if I did. In my mind I was comparing it to php where there is a simple function that converts local time to UTC.

I will admit that I didn't realise the original question was about v1 so for that I also apologise.

There is probably a simple function to convert local time to UTC in most languages. That doesn't mean that changing a large, complicated piece of software to use it as an option, and then validating that it all still works correctly, is either simple or quick.