Old Meteoplug Forum - now read-only!

Hello

I upload to Meteoplug using a Meteobridge TL-MR3020 server very 10 mins, and live every 30 secs. Once a day, Meteoplug processing on the upload stops. I think continues to be uploaded, but graphing stops. Data sent to WU continues fine. To get Meteoplug to start processing again, I need to reboot the Meteobridge server.

Just before the processing stops, the client log shows

logger (18.02.2014 23:02:58): cannot open pipe "gzip >/var/run/meteohub/inetupload.z": Cannot allocate memory
logger (18.02.2014 23:51:37): open named pipe "/dev/meteobridge" failed: No such device or address

Please advise.

That looks like shortage of RAM on the Meteobridge. What upload service do you have running there and what kind of weather station is it?

AcuRIte 01050C/01055 sensors report to AcuRite Bridge. The Bridge is connected by ethernet to TP-LINK TL-MR3020 flashed as a Meteobridge, 2.1. The TL-MR3020 connects to my LAN via wi-fi. Meteobridge uploads to both WU and Meteoplug every 10 minutes, along with real-time data every 10 seconds. The AcuRite Bridge also reports to the Acu-Link site every 10 minutes.

When processing stops at Meteoplug, WU continues its 10 min update and live update.

when yo send the MAC of your meteobridge to "info(at)meteobridge.com" I might be able to login and understand what makes storage that tight.

Sent

Hello again -

Still having the problem: Meteoplug is loosing data after about 20 - 22 hrs of service. Realtime data continues but 10 min upload graphing data stops. I'm using a Meteobridge (see above) but it's running with only 5% memory free. I don't know why.

One other point: The setup of Meteobridge server ran with Meteoplug without a problem for the month of introductory time. When I subscribed to Meteoplug the problem started within a day.

Suggestions? I've re-flashed the Metebridge once. Could I re-set it to factory defaults and try it again? Anything?

m

I don't think you can do much about it. I will investigate during the weekend how to get this solved.
It ocurs with Acurite bridge only as Meteobridge needs to monitor TCP/IP traffic and the library
needed for that is extremely huge (in terms of what the Meteobridge can offer).

As Meteoplug upload service stores data locally for some time so that an upload interruption
of a few hours can be compensated without loosing data, there seems to happen the out of
memory sitation when this buffer filly up. Most easy way to solve would be to reduce Meteoplug
data buffer, but I would like to check if I can reduce memory footprint in another way.

Reflashing will not cure. Reset to factory defaults can be done by pressing reboot button
and keep pressed until fifth LED blink (please also see "getting started" section of wiki, near the
end of the page)

I found some time to look at this and optimized storage use a bit. Please do a reboot of your Meteobridge to get the new code. Storage consumption display on "system page" should now show sligthly more friendly values. Hopefully, nothing else is impacted negatively by that... we 'll see.

Hello -

Thanks for the explanation. That helps me understand the problem.

I rebooted and was updated to 3345. The system panel shows a 2% increase in free RAM, about 4k. Not a lot, but it might be enough. I'll let you know.

Platform: TP-LINK TL-MR3020
RAM: 29364 kB total, 1976 kB free (93% used)
Software Version: Linux/OpenWrt 3.3.8 mips (BE), FW 1.2, MeteoBridge 2.1 (3345)
Version Control: on next reboot
Uptime: 10 hours, 8 minutes Buffer Usage: 0 items (0%)

If theres's anything I can do on this end to reduce the load on this end, let me know. For instance, I'm running the Acurite 5 in 1 sensor + two extra temp/humid sensors. I'm sending data from the two sensors to Meteoplug (Shed and Greenhouse temps), but I could try removing them to see if the load changes.

- m

2% is about 600k in my math, which should help. Please let me know if the problem is solved by that.