Extreme values when data missing.

[BCJKiwi]

Was testing some alternative configs on hardware.
shut down all connections to CumulusMX and then the engine.
Disconnected the station.

A short while later ~ 10 mins, reconnected everything and waited for the VUE to reconnect to the ISS while having a cuppa.
Then started the engine again.

Now, some time later looking at the charts there are some funny numbers.
So here I am reporting what I see in the spirit of Beta testing and not making any sort of complaint or looking for any fix as a station should be running all the time - right? Just something for Steve (and Mark?) to think about when they get some spare time - maybe next year??

Dewpoint drops to zero and comes back,
Apparent temp dives to -83.8,
Indoor temp takes a direct line across the missing data,
Wind Chill drops to -18,
Outside temp drops to zero,
Pressure appears to take a direct line same as inside temp,
There is no wind as the VUE is inside for these beta tests but wind speed jumps from 0 to 410,
Wind dir has a gap
Indoor humidity takes a direct line across the missing data,
Outdoor Humidity drops to zero,
No rain either and this displays a gap.

Don't know if this is what is coming from the station but would nice to have;
a consistent treatment of missing data, either a straight line across the gap, or, a gap.
some sort of sanity check on first data after start which rejects it if too far off the last value.

Outside and Apparent temp gauges also behaved strangely on first open of the page. Both show a scale from -100 to 40. Every 10 secs, both gauges reset to -100 and wind up to somewhere near current values.
Humidity and pressure behaved normally.
As all values are relatively steady as the VUE is inside, it is difficult to see what is happening. However the temps did keep moving very slowly (by 0.1 degree) every few minutes.
Page refreshes did not make any difference. Navigating away from the page and coming back also made no difference.

A new normal upload of data made no difference.

Restarted the engine and the gauge behaviour continued from the temps it was on before the restart.

[BCJKiwi]

OK I'm back ~ an hour later.

Gauges forecast displays,
Page Auto-update limit reached. Click the status led to continue. Clicked LED and gauge behaviour continues as before.

Went to look at MXdiags and saw that the Cumulus restart has some unusual info;

2015-01-18 18:40:23.002 Done uploading graph data files
Key pressed - terminating
2015-01-18 18:44:03.322 Cumulus closing
2015-01-18 18:44:03.337 Writing Cumulus.ini file
2015-01-18 18:44:03.365 Closing connection
2015-01-18 18:44:03.482 System.InvalidOperationException: The port is closed.
at System.IO.Ports.SerialPort.get_BytesToRead()
at .(Int32 )
2015-01-18 18:44:03.483 Station stopped

The new log starts;

2015-01-18 18:44:13.442 ========================== Cumulus MX starting ==========================
2015-01-18 18:44:13.442 Cumulus MX v.3.0.0 build 3009

and continues normally.

So the gauges have been playing catch up now for over an hour and continuing.

[steve]

There are some 'out of range' values in the archive data that I'm not currently checking, I'll have a look at doing that. Note that 'default' values will still be logged until the first valid data is read, but that's not a new issue. Maybe next year...

[BCJKiwi]

The apparent temp gauge is still catching up.
Now gone midnight so some 5.5 hrs after this started.

Of course it kept stopping while I was not in front of the PC.
Reset counter to 4 secs and pageupdatelimit to 120 to speed things up.
The Temp gauge is now caught up but still waiting for the Apparent temp gauge.

[steve]

I don't know what you mean by 'catching up'. If I look at your apparent temp gauge now, it shows 28.4C, the same as your 'now' page.

[BCJKiwi]

What it looks like from the perspective of the user - no idea what is going on in the code - after the restart described in the first post, it appears that the temp and the apparent temp gauge started at the beginning of whatever data source they read and stepped through every entry until they finally reached the current time.

Initially the default settings in scripts\gauges.js settings;
counter : 10,
and
pageUpdateLimit : 20,
were in play. Only the Temp and App temp (and other values on that gauge) behaved as described in the first post. The gauges would have cycled from -100 to current value some 360 times per hour for some 4 to 4.5 hours at which time I changed the counter to 4 secs and the update limit to 120 to speed the process up. It did not finish doing this until around midnight - I wasn't watching at the time - another ~ 900 times.
This process was not in a loop re-reading the same data point as the current values were changing. The temperature gauge stopped a little before midnight and I suspect the apparent gauge stopped at midnight as I understand the gauges tool tips only display highs and lows for the current day since midnight - if not then it stopped a short while after midnight.

Hope this helps.