Periodic wild temperature readings **solved**

[guineau1]

Platform: TL-MR3020 (no USB hub)
RAM: 29364 kB total, 2952 kB free (89% used)
SW Version: Meteobridge 3.1 (Dec 22 2016, build 10776), FW 1.4

WS-1400-IP weather station

I've had a Meteobridge setup for a couple months. I registered and paid for a license.

Every so often (several days to couple weeks) the Meteobridge seems to flake out and return very high temperature readings. This is similar to the behavior when the initial trial period ran out. Rebooting the device clears it up, but it comes back. I know it is the meteobridge since the station console and the Observer-IP both report correct temperatures while the meteobrigde is reporting, for example, 138.2F.

Here is a screenshot of my station KCASANJO392 on Weather Underground:

Crazy temps from Meteobridge on WUnderground

meteobridge_flakeout.png (110.08 KiB) Viewed 3693 times

Any help greatly appreciated as this has "polluted" my data on Weather Underground!!

john

[guineau1]

Unfortunately after a reboot (a couple times) this time it does not seem to fix the issue. Is my Meteobridge corrupted somehow and need to be reflashed?

Underground data

meteobridge_flakeout_WUG.png (44.39 KiB) Viewed 3689 times

Here is the Meteobridge Live Data:

Live Data screenshot

meteobridge_flakeout_Live_Data.jpg (199.22 KiB) Viewed 3689 times

[guineau1]

After poking around inside Meteobridge (ssh root@<meteobridge>) I found that it looks like the Meteobridge is just using HTTP to grab the Station and LiveData web pages from the Observer-IP. It must then parse out the raw data.

Turns out the the Observer-IP itself gets reaaaal slow over time. Directly viewing the two pages that Meteobridge reads shows that those pages are very slow to update. often updating only partially.

Rebooting the Observer-IP seems to solve the problem. I think in the past couple times, I rebooted them both. Looks like Observer-IP is the real culprit!

Hopefully this will help future users...

john

[admin]

Your assumption is right, MB reads data from the Observer-IP web interface.
But even when this renders unresponsive, MB should not report fake data.
I will have a look at the code. Thanks for analyzing the root cause!

[guineau1]

Just came across this:

https://github.com/poblabs/weewx-ObserverIP

Which might be helpful...

[admin]

Thanks again for going into detail here. Much appreciated.

The articled linked above is very interesting but imho not well suited to
solve the problem for regular plug&play users. Redirecting all traffic through the Meteobridge
will need to give the Meteobridge a fixed IP in the lAN and the Observer-IP would inoperational
when Meteobridge is not running. That will add too much interdependencies and make things
even more complex.

What I would like to do is to ignore data from stalled pages and when a stall occurs to give the
Observer IP some time to rest before start polling data again. I think this will be a less
invasive way to deal with the issue.

Would you allow me to login to your Meteobridge, so that I can find out how HTML response looks like,
when the Observer IP stalls? I you could send remote URL and password to info(at)smartbedded.com,
that would be very helpful.

[guineau1]

Totally understand on 'plug and play' for basic users. I think that article took the hard way around - why not just point the Observer-IP directly to the Meteobridge and let Meteobridge "pretend" to be Underground (serve the correct pages as the article alludes to). Setting up forwarding is definitely not for the average user!

I'll set up remote access and get you the info. To force it to hang quicker, I thin (from that other article) that you just need to make Metoebridge poll the Observer-IP more frequently. I'll give that a try and see if I can make it fail.

john

[guineau1]

Hmm, I setup up Meteobridge to post to WUnderground every 5 seconds (that used to break the ObserverIP more quickly so I was using 10), and set up the remote login and rebooted. Discovered I have a new FW version 3.2 from 1/21/2017:

SW Version: Meteobridge 3.2 (Jan 21 2017, build 10817), FW 1.4

So far I have not had any issues with ObserverIP hanging up and Meteobridge posting wild temps. I do see occasional new messages:

logger (21.01.2017 18:36:55): error while waiting for data from weather stations
logger (21.01.2017 18:36:55): disconnect station 0 (Observer-IP): wrap-up
logger (21.01.2017 18:36:55): data logger stopped.
logger (21.01.2017 18:36:55): Historical data loaded (23 records).
logger (21.01.2017 18:36:55): data logger (version 5.0t, build 10817) started.
logger (21.01.2017 18:36:56): connect station 0 (Observer-IP via USB HID).
logger (21.01.2017 18:36:57): Waiting for sensors (wind0, rain0, th0, thb0) providing data. Services paused.
logger (21.01.2017 18:37:02): edited value for parameter "thb0dew amax" applied: 13.33
logger (21.01.2017 18:37:32): edited value for parameter "th0heatindex amax" applied: 78.67
logger (21.01.2017 18:37:56): Give up waiting for sensors (wind0, rain0, th0, thb0) providing data (waited 60 secs). Services started.

So I'm guessing the Meteobridge guys fixed something...

Also, I saw in tanopther post that someone had discovered that the "wild" temperatures were in fact Meteobridge interpreting degrees F as degrees C. That made it easy to fix all my max temps for historical data.

I'll keep the forum on how things go...

john

[guineau1]

So far so good. I think the latest firmware must have resolved the ObserverIP issue!