Migration underway

[Darmain]

I have been running Cumulus since late 2011, which has been running almost continuously apart from about a month in 2012 when we moved house. The station is a Davis Vantage Pro2 with Solar and UV. The current PC is a Fujitsu Siemens Pentium 4 2.5GHz, running Windows XP, sat behind a multi-layered firewall, feeding my website, WOW and WU. I have three of the machines, all cast off from work. The first died of PSU failure. This is the second, but has been occasionally switched to the third, to save the wear. I have the steel series gauges running on the website, which are just brilliant.

I am working on a project to create a wall mounted, multi-scaled analogue weather gauge using a stepper motor, controlled by a raspberry pi. This reads the JSON file from the website and provides a facility where you can use buttons to select if the gauge reads temperature, pressure, wind speed or wind direction. The plan is also to have a small LCD display to provide other information. I have one mate who kindly created the scale legend on AutoCAD and another mate is going to CNC machine the wooden back plate to hold all this together, look the part and be able to hang on the wall. This is a project in progress. I have the first prototype built using a RP2, on breadboards, using a mocked up analogue display and wires everywhere, all working nicely. The application is coded in Python3. I plan to start the second prototype, which moves over to a PI Zero and tidies things up a bit. Once this is done I hope to get on and design the final packaging.

Anyway, before this I aim to retire the PC and use a RP2, with an external USB HDD to run Cumulus. The equipment lives in my back bedroom, which, if the door is left closed, heats up nicely in the day due to the power consumption of the PC. It is this power consumption that I aim to reduce. I would have done this before but the problem was quite simply that I could find no other application that does anywhere near as good a job as Cumulus. I see there are ways to run x86 shells on the PI, then run WINE, then run Cumulus. I'm not sure "run"would be the word, more like hobble I suspect. I didn't bother trying as I guessed it would very likely be unusable. The discovery of CumulusMX is just outstanding, and another Donation is on its way. Sure there maybe a few features to contend with but I am happy to help with this beta testing. Raspberry PI, here we come.

Many thanks,
Dave

[Darmain]

Migration is almost there, apart from a problem with the Weather underground ( https://cumulus.hosiene.co.uk/viewtopic.php?f=27&t=15757). Have noticed that if you monitor the system using the RPi browser, pointing to localhost then the terminal service gets a bit wobbly and occasional falls apart. Monitor from another machine and it seems stable.

[Darmain]

Good news, the system stayed up and stable all night. WEather underground is also alive following the correction to the password formatting.

For the first time for a long time it was "quiet" during the night, what with no fans running! .

There was an odd one today that I caught by chance. The steel series gauges on my web site suddenly stopped working as the hour passed 12:00. I tried reloading but it was clear that the file wasn't locked, it appeared to be invalid to the client side script. The complaint was that the weather station went off line and had been off line for 270 odd days. The system was most definitely not off line and clearly I could access the website. I jumped on my FTP client, logged into the website and downloaded the realtimeguages.txt file for analysis. Not immediately obvious whats going on here but I can't see anything wrong. Then several minutes later it came back to life, so I grabbed that file as well. I attach both files for gremlin analysis.

Code: Select all

{"date":"12:03",
"temp":"10.1",
"tempTL":"4.8",
"tempTH":"10.1",
"intemp":"17.6",
"dew":"8.2",
"dewpointTL":"3.9",
"dewpointTH":"8.2",
"apptemp":"8.0",
"apptempTL":"2.5",
"apptempTH":"8.0",
"wchill":"10.1",
"wchillTL":"3.6",
"heatindex":"10.1",
"heatindexTH":"10.1",
"humidex":"10.5",
"wlatest":"2.4",
"wspeed":"2.4",
"wgust":"6.2",
"wgustTM":"7.1",
"bearing":"105",
"avgbearing":"73",
"press":"1008.26",
"pressTL":"1008.26",
"pressTH":"1011.07",
"pressL":"966.78",
"pressH":"1040.40",
"rfall":"0.0",
"rrate":"0.0",
"rrateTM":"0.0",
"hum":"88",
"humTL":"88",
"humTH":"96",
"inhum":"70",
"SensorContactLost":"0",
"forecast":"Unsettled, rain later",
"tempunit":"C",
"windunit":"m/s",
"pressunit":"mb",
"rainunit":"mm",
"temptrend":"+0.9",
"TtempTL":"01:58",
"TtempTH":"12:02",
"TdewpointTL":"01:58",
"TdewpointTH":"12:02",
"TapptempTL":"02:34",
"TapptempTH":"12:02",
"TwchillTL":"02:34",
"TheatindexTH":"12:02",
"TrrateTM":"00:00",
"ThourlyrainTH":"00:00",
"LastRainTipISO":"2016-10-19 16:18",
"hourlyrainTH":"0.0",
"ThumTL":"11:50",
"ThumTH":"06:15",
"TpressTL":"11:59",
"TpressTH":"00:00",
"presstrendval":"-0.26",
"Tbeaufort":"F2",
"TwgustTM":"11:44",
"windTM":"2.8",
"bearingTM":"57",
"timeUTC":"2016,3,23,11,3,8",
"BearingRangeFrom10":"360",
"BearingRangeTo10":"130",
"UV":"0.6",
"UVTH":"0.7",
"SolarRad":"225",
"SolarTM":"225",
"CurrentSolarMax":"378",
"domwinddir":"NE",
"WindRoseData":[23.7,159.7,1833.4,3719.8,1978.8,218.0,6.2,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0],
"windrun":"56.6",
"version":"3.0.0",
"build":"3041",
"ver":"11"}

Code: Select all

{"date":"12:12",
"temp":"10.3",
"tempTL":"4.8",
"tempTH":"10.3",
"intemp":"17.7",
"dew":"8.3",
"dewpointTL":"3.9",
"dewpointTH":"8.3",
"apptemp":"7.6",
"apptempTL":"2.5",
"apptempTH":"8.1",
"wchill":"10.3",
"wchillTL":"3.6",
"heatindex":"10.3",
"heatindexTH":"10.3",
"humidex":"10.8",
"wlatest":"2.8",
"wspeed":"3.3",
"wgust":"8.5",
"wgustTM":"8.5",
"bearing":"102",
"avgbearing":"77",
"press":"1008.26",
"pressTL":"1008.26",
"pressTH":"1011.07",
"pressL":"966.78",
"pressH":"1040.40",
"rfall":"0.0",
"rrate":"0.0",
"rrateTM":"0.0",
"hum":"87",
"humTL":"87",
"humTH":"96",
"inhum":"70",
"SensorContactLost":"0",
"forecast":"Unsettled, rain later",
"tempunit":"C",
"windunit":"m/s",
"pressunit":"mb",
"rainunit":"mm",
"temptrend":"+0.9",
"TtempTL":"01:58",
"TtempTH":"12:11",
"TdewpointTL":"01:58",
"TdewpointTH":"12:11",
"TapptempTL":"02:34",
"TapptempTH":"12:06",
"TwchillTL":"02:34",
"TheatindexTH":"12:11",
"TrrateTM":"00:00",
"ThourlyrainTH":"00:00",
"LastRainTipISO":"2016-10-19 16:18",
"hourlyrainTH":"0.0",
"ThumTL":"12:04",
"ThumTH":"06:15",
"TpressTL":"11:59",
"TpressTH":"00:00",
"presstrendval":"-0.26",
"Tbeaufort":"F2",
"TwgustTM":"12:05",
"windTM":"3.3",
"bearingTM":"46",
"timeUTC":"2016,12,23,11,12,8",
"BearingRangeFrom10":"020",
"BearingRangeTo10":"160",
"UV":"0.7",
"UVTH":"0.7",
"SolarRad":"214",
"SolarTM":"225",
"CurrentSolarMax":"384",
"domwinddir":"NE",
"WindRoseData":[23.7,187.2,1915.4,3651.1,2111.1,284.3,14.2,4.7,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0],
"windrun":"58.0",
"version":"3.0.0",
"build":"3041",
"ver":"11"}

[steve]

"timeUTC":"2016,3,23,11,3,8",

Looks like the clock on your Pi went wrong for a while.

[mcrossley]

Then "timeUTC":"2016,12,23,11,12,8"

The template for the json files isn't using minutes instead of the month is it?

[steve]

Then "timeUTC":"2016,12,23,11,12,8"

The template for the json files isn't using minutes instead of the month is it?

Good catch - probably using the realtimegaugesT.txt from Cumulus 1 rather than MX.

[Darmain]

Ah, well reasoned. You are right, I overwrote the "web" folder from the Cumulus 1 installation as part of the data transfer. I have corrected this by recalling the MX version of realtimegaugesT.txt, which I can see is indeed different.

Why did I do this? Well, I've tweeked the web templates so I over wrote them to preserve this. It appeared on the face of it that this worked ok. Apparently not quite.

The question is, is there anything else that I have broken?

Another question, if I may. The original steel series has mouse over graphs, which the MX system does not provide. Has the java scripts been updated to fix this? I can have a look and fix this myself but its silly if this has already been done.

The system is still operating in a stable condition. The only adviser is to not use the RPi web browser to access the interface, as this seems to occasionally take the service down. I am using the Chromium web browser on my Ubuntu PC to access the interface and this works very well.

Many thanks once again.
Dave

[steve]

The question is, is there anything else that I have broken?

It would probably be easier to answer that after seeing your web site. Any custom date and time formats you may have used will need changing as in the MX instructions. The graphs page is now obviously completely different.

Another question, if I may. The original steel series has mouse over graphs, which the MX system does not provide. Has the java scripts been updated to fix this? I can have a look and fix this myself but its silly if this has already been done.

See this 'sticky' thread - https://cumulus.hosiene.co.uk/viewtopic.php?f=27&t=13189

[Darmain]

I have spent this evening working on the web output and concluded that porting the Cumulus 1 arrangements over to Cumulus-MX was not the way to go. So, I took down the original website and migrated it to the Cumulus-MX settings and scripts. After several false starts I managed to get the configurations all working. Then I looked at the resulting website...

Wow, what a difference!! This is absolutely fantastic work and a major upgrade over Cumulus 1, which was very good in itself to be fair!

I have backed everything up and now let the PI run.

[Adrian Hudson]

Hi Dave,

Could you explain and expand on what you mean by this please: "concluded that porting the Cumulus 1 arrangements over to Cumulus-MX was not the way to go"

I have had a couple of attempts to mode to the Pi over the years - I don't run the default web site which is probably half my problem but I am interested to read this thread and see how you get on.

Thanks!

[steve]

I'm interested to hear what Dave means by that too. The standard MX web site is not significantly different to the standard Cumulus 1 web site, with the exceptions of the gauges and graphs, the standard versions of which are now the SteelSeries gauges and Highcharts graphs developed by Mark Crossley which were already available for Cumulus 1.

Any custom web site which worked with Cumulus 1 should work with little change, the main issues being where any date/time modifiers were used in web tags, and these are simple enough to change, and the lack of any static images in MX, e.g. for graphs. Custom pages for the gauges (if SteelSeries weren't already being used) and the graphs are where the biggest changes would be required.

[Darmain]

Hi Chaps,

Please don't take my comment as anyway towards criticism, on the contrary as you will see. Quite simply what I meant was I had considered hacking away at the HTML to make the Cumulus 1 web pages work correctly with Cumulus MX. It was quite clear that the original graphs were not supported and I assumed there would be no graphs at all. I then realised that this would mean that all the web page templates would need editing to remove the "Trends" link. In addition I didn't know how much work would be needed to get the time / date accesses right. Finally there was the "anything else" factor of things I didn't know that maybe wrong.

So I decided that the better way was to drop the original web templates and to replace it entirely with the new versions from Cumulus MX. This proved to be highly successful. First the Steel-series was straight in AND improved. Secondly the graphs were not removed at all and were also much improved, actually in a different league. Therefore there is much to be gained by switching to the new templates and then modifying them to suit your own style, as necessary.

The reason I actually changed them in the first place was because my website was being reverse proxied into the main website of Wessex 4x4 response. I provide a local, extreme weather forecasting service for the Wiltshire, Dorset and Somerset counties, in order to give heads up alerts to our responders. We have a couple of other responders that also run PWS' so our web master linked them all into the main website for all to see. The issue was the CSS was different and it stood out like a sore thumb, so I changed it to match the design of the main site.

Now the web master has moved on and a new guy sits in control. As a result the W4x4R website is due to be totally replaced and I doubt the weather stations will be included. Therefore I am very tempted to leave my site as it is.

I hope that all makes sense.

[Darmain]

Hi.

I've got an odd question (Hey, its me!! ). I wanted to recall the average temperature for a given month. I know you could do that on C-1. Where is this on C-MX?

The RPi is stable and all is well. An upgrade to the hardware was the addition of a UPS for the PI, the HDD, the ethernet switch and the ADSL router. I know from bitter experience that crashing a linux system without unmounting the disks can go bad. This also adds the benefit of power conditioning.

Many thanks,

Dave

[steve]

I've got an odd question (Hey, its me!! ). I wanted to recall the average temperature for a given month. I know you could do that on C-1. Where is this on C-MX?

Not written yet, along with a number of other facilities, listed in a 'sticky' post somewhere.

[Darmain]

Ah, okay Steve. No worries. I did search the forum for this but obviously didn't pick the right words.
Many thanks.