Meteohub Update 4.0 "multi station" (highly experimental)

[admin]

With release 4.0 Meteohub goes another big step by adding support for handling more than one weather station at a time. This multi weather station feature is quite unique and made a major rework of the data logging architecture of Meteohub necessary.

This release candidate will still have bugs in it, so you should not consider installing it in a production like environment. However, if you have more than one weather station and you are interested to make these avail to Meteohub, this release might be worth a try.

Multi-station feature can be used in demo mode and with standard licenses until end of March.

Your feedback on this release is highly welcome. Being able to process the data of more than one station provides a lot of new options. You might think of fail-over sensors, evaluations of measurement differences between sensors of different stations, new type of difference graphs, etc. All this is not implemented now, but Meteohub now provides a base to make this happen soon.

Another side effect of the multi station feature is, that Meteohub supports weather station type named "Plug-In", where alien data logging programs can be used for feeding weather data into Meteohub. This allows the experienced user to write a simple decoding program for his preferred weather station (1-wire, for example) that is not yet directly supported by Meteohub. Manual version 4.0 explains how that works in detail. It is quite easy and straight forward.

I hope you enjoy the new release... and please accept it having still some bugs. I had to change so much code, so lots of chances for me to introduce new bugs... and usually I don't void such a chance ;-)

Update:
http://www.meteohub.de/joomla/index.php ... &Itemid=29

NSLU2 USB Stick Image 4GB:
http://www.meteohub.de/joomla/index.php ... &Itemid=29

NSLU2 USB Stick Image 2GB:
http://www.meteohub.de/joomla/index.php ... &Itemid=29

x86 Installation Image:
http://www.meteohub.de/joomla/index.php ... &Itemid=29

VMware Image:
http://www.meteohub.de/joomla/index.php ... &Itemid=29

Documentation:
English: http://www.meteohub.de/joomla/index.php ... &Itemid=29
German: http://www.meteohub.de/joomla/index.php ... &Itemid=29

[skyewright]

Another side effect of the multi station feature is, that Meteohub supports weather station type named "Plug-In", where alien data logging programs can be used for feeding weather data into Meteohub....It is quite easy and straight forward.

Excellent. An elegantly simple solution, again. :cheer:

v4.0 now running on an ALIX1.D, connected to an IP-RFXCOM.

I presume that is wasn't intended that it should automatically 'adopt' the v3.0 single weather station as the first multiple station?

A few things I've noticed so far
The WMR200 outdoor sensor appears as THGR810 (it's THGN801 on v3)

The WMR928 rain gauge (STR918 in v3) is not being picked up.

None of my THC238 (temp only, seen as THWR288 in v3) sensors are being picked up.

Meteohub log reports lots of rfxcom related errors which probably relate to the THWR288 problem at least. e.g.

ogger (31.01.2009 21:41:24): wrong rfxcom data in record. Nibble a is not decimal.
logger (31.01.2009 21:41:24): wrong rfxcom data in record. Nibble a is not decimal.
logger (31.01.2009 21:41:29): station 0 (RFXCOM), wrong checksum (0f vs computed 3c) for sensor model ea4c (THWR288, THN132N) in byte sequence: 50 ea 4c 20 ae 00 13 c0 43 05 00
logger (31.01.2009 21:41:29): station 0 (RFXCOM), wrong checksum (0f vs computed 3c) for sensor model ea4c (THWR288, THN132N) in byte sequence: 50 ea 4c 20 ae 00 13 c0 43 05 00
logger (31.01.2009 21:41:31): station 0 (RFXCOM), wrong checksum (06 vs computed 33) for sensor model ea4c (THWR288, THN132N) in byte sequence: 50 ea 4c 10 91 40 06 30 f3 0d 00

Maybe tomorrow I'll try attaching one of my consoles (probably WMR200) direct, as a 'second' station?

If you want remote access just let me know (the details you already have route to the other Meteohub, so I'll need to swap router settings).

[skyewright]

...and I just noticed that the THGR810 (which maybe should be a THGN801? see above) has the same # as another sensor (a THGR228N).
Not sure if that is intended to be possible or not? They both seem to be working fine, but I thought maybe I should mention it...

[admin]

thanks for your observations. Having a chance to login to your system, will certainly help to get things sorted out fast.

I am not sure if RFXCOM has ever reported THGN801. in WMR200 mode it has done this, and also does it in v4.0.

RFXCOM sensors can have the same original channel ID without having problems as long as the sensors can be differentiated by type. This seems to be the case here, so no problem. If two identical sensors get the very same channel ID after reset or battery change, we are in trouble, as Meteohub cannot keep the data separate. This would result in one recognized sensor with flipping values.

STR918 is a wind sensor and is being picked up here at my USB RFXCOM, rain gauges are RGR682/RGR918 which all have the same identification pattern, so Meteohub uses the first matching name for display, RGR682. I will change that to RGR918 to better match standard situation.

I will have a closer look at the pure temp sensor problem.

[admin]

uv sensor and pure temp sensor are impacted by a checksum bug when in RFXCOM mode. WIll be fixed with next update.

[skyewright]

Having a chance to login to your system, will certainly help to get things sorted out fast.

The details you last used should now get you to meteohub-c the Meteohub being used for this test. If you need the details again, please e-mail.

I am not sure if RFXCOM has ever reported THGN801.

?
It is listed and illustrated in Appendix G, and that's what my WMR200 outdoor as always reported as with v3 & earlier (USB-RFXCOM & IP-RFXCOM).
I think the THGR810 (in a separate cell in Appendix G) is an "extra" sensor?

If two identical sensors get the very same channel ID after reset.

No sensor reset involved here, just the switch to v4.
Since I run 6x WMR928 "3 channel" sensors I'm familiar with sensor conflicts, but in approaching a year of running I've not known the THGN801 (WMR200 outdoor) to conflict with anything.

STR918 is a wind sensor and is being picked up here at my USB RFXCOM, rain gauges are RGR682/RGR918 which all have the same identification pattern

Opps. Copy & Paste error - I should have just typed!
It is the WMR928 rain gauge (which appears a RGR126 in v3) which is not appearing in v4.

[skyewright]

Since I run 6x WMR928 "3 channel" sensors I'm familiar with sensor conflicts, but in approaching a year of running I've not known the THGN801 (WMR200 outdoor) to conflict with anything.

Thought:
Since the WMR200 outdoor transmits every ~60 seconds and WMR928 "extra" sensors transmit every ~45seconds the odd conflict now and then would not be noticed since the 'next' signal would not conflict.
Maybe that explains why I've never seen a conflict between the WMR928 and WMR200 outdoors that I'd been expecting? :)

PS. V4 now connected to IP-RFXCOM, and to meteohub-b (running v3 and connected to a USB-RFXCOM) via port 5557 (which I don't think is a documented option in the v4 manual - that just mentions 5500, which I think assumes v4 on the other Meteohub?). So meteohub-c is now seeing each sensor via two different routes. :)
Same sensor id's and ommisions in both cases - as you'd presumably expect?

[skyewright]

None of my THC238 (temp only, seen as THWR288 in v3) sensors are being picked up.

A 'visit' from Boris has fixed this (thanks :) ), so OS temp only should be fine for all in the next release.

[skyewright]

and...

The calibrations on the Settings page don't seem to be being applied?
I have tried clicking "Save" on the settings page.

PS. Now attached to 3 stations.
IP-RFXCOM (direct)
USB-RFXCOM via meteohub-b:5557
WMR200 via USB HID

WMR200 reports 3 sensors:
WGR800 - same data as WGR800 via the other two
THGN801 - same data as for 'THGR810' via the other 2
'unknown' - The WMR200 indoor temp/hum/baro (not available to RFXCOM, of course).
The WMR200 rain gauge isn't getting through, but that isn't a surprise as the console is showing --- anyway (i.e. in its current position by the ALIX 1D the console can't 'hear' the rain gauge).

At least during recomputation the system seems to be having difficulty coping (i.e. missing some sensor readings) with the (probably unrealistic) load I'm giving it. I don't think I ever recall seeing System Load figures as high as 5 on the ALIX 1D before.
Stress testing, eh! ;)
I'll see how it settles down.

Edit: Peak load seen so far 8.78!

[skyewright]

The WMR928 rain gauge is not being picked up.

WMR928 rain gauge (RGR126) is now being picked up. :cheer:

[admin]

I could fix a number of bugs in the data logger that showed up on skyewright's setup. The download is now updated to update-v4.0-rc2.upd.

fixes are:
- rgr126 recognition
- pure temp sensor recognition
- wmr200 thb sensor fix
- dropping connection to tcpip connected sources solved. Should be ready to handle even extrem stress now.

If you downloaded before this messages was written you are still on "rc1", please download the "rc2"-update and install.

[skyewright]

- dropping connection to tcpip connected sources solved. Should be ready to handle even extrem stress now.

System load currently 8.65 (just started another recomputation) and all sensors by all routes are being picked up. That's 29 sensors - it would be 30 if the WMR200 rain gauge had a better transmitter!

PS. For late arrivals I don't really have 30 sensors!

PPS. I do have a few 1-wire sensors and a 1-wire USB adaptor, so if someone has a suitable bit of code handy I could try applying even more pressure (and test the plug-in interface).
I think I could write some myself (and may well do in due course), but spare time is a bit short at present...

[skyewright]

System load currently 8.65 (just started another recomputation) and all sensors by all routes are being picked up.

System load peaked at over 9, and all sensors were still registering. :)

It seems that with one thing and another a whole load of recomputations were running simultaneously.

Recomputation now shows 'completed' and the system load has settled to well below 1 (0.05 when I just looked!).

Looking through the meteohub log there are 10 recomputation of weather data finished: messages since the last recomputation of weather data started.

It would be real nice if either:
a) A fresh recomputation could 'kill' any already in progress,
or
b) A fresh recomputation could set some sort of flag that could be seen by earlier recomputations causing them to stop themselves when they next checked the flag value.

Just an idea...

[piombo]

I'm having some problems with the new version and my WRM200.

1) There is no extra UV-Sensor any more (used to be uv0 before update). It doesn't show up at the sensor-page. Standard senors work.
2) At the dashboard-setting-page there the wind-sensor has no name and shows wind(). If I chose wind(0) out of a lot of wind0 () meteohub will not change it for the next time (dashboard works, even wind).
3) wind chill (calculated by meteohub) uses the inside-temperature
4) Graphics: Air pressure is no displayed at sea pressure-level any more (like it used to be before). It now uses the local pressure.

Thanks for help.

[skyewright]

3) wind chill (calculated by meteohub) uses the inside-temperature
4) Graphics: Air pressure is no displayed at sea pressure-level any more (like it used to be before). It now uses the local pressure.

What do you have set in Weather Station:

Station's Altitude
Sea Level calculation
Wind Chill calculation