Cumulus problem after system time change

[gluepack]

It may be coincidence (the system time change) but...

I have been running on an old laptop under XT for a while now, since service guy screwed up my desktop Win8 system.
I have noticed that the system clock has been slowing for a while and was about four minutes out so I decided to restart the system.
I closed Cumulus, restarted the system, changed the clock and then restarted Cumulus.
It went through a reading from memory process going from FFFFFF? down to 0 and then went into a Processing Data phase finally crashing. I have attached the bugreport.txt file.
I restarted Cumulus as suggested and it came up ok apart from the fact that now the data is screwed e.g. today's low -35C, high gust 198 km/h, at least.

[steve]

It crashed in a third-party routine which calculates dew point; presumably due to garbage data from the station (perhaps it was reading invalid entries). The clock change could be relevant because it would control how many entries it thought it needed to download to catch up. If you zip up the diags folder and attach it, I'll take a look.

[gluepack]

Lol, now, for no apparent reason, Cumulus is continually reporting lost contact with the sensor. However, having restarted Cumulus again that isn't occurring.

I have attached the zipped folder.

The only bad reading carried forward to Wunderground (currently) appears to be the high gust.

[steve]

It was reading invalid data because the first block of memory had mostly FF in it:

9/3/2014 11:05:39.859 : EWUSB: EW startup data line 01 55 AA FF FF FF FF FF FF
9/3/2014 11:05:39.859 : EWUSB: EW startup data line 02 FF FF FF FF FF FF FF FF
9/3/2014 11:05:39.875 : EWUSB: EW startup data line 03 FF FF FF FF FF FF FF FF
9/3/2014 11:05:39.875 : EWUSB: EW startup data line 04 FF FF FF FF FF FF FF FF

The current data pointer is in there, so it started reading at FFFF, which is not a good place to start, so it read invalid data. When you restarted it, the first block was OK:

9/3/2014 11:16:51.984 : EWUSB: EW startup data line 01 55 AA FF FF FF FF FF FF
9/3/2014 11:16:52.000 : EWUSB: EW startup data line 02 FF FF FF FF FF FF FF FF
9/3/2014 11:16:52.046 : EWUSB: EW startup data line 03 05 20 02 81 11 00 00 00
9/3/2014 11:16:52.046 : EWUSB: EW startup data line 04 01 00 00 F0 0F 00 10 C5

If you stop Cumulus and find the backup files that it took when you started it up at 1105:

9/3/2014 11:05:31.000 : Creating backup folder C:\Cumulus\backup\20140903110530\

and copy those files into the data folder, overwriting the ones that are there, then when you start it up again it should sort itself out. If you delete the bad entries from WU before you start it up again, it should replace them with sensible ones, assuming you have 'catch up' set for WU.

[gluepack]

Thanks for that.

It may be ok now ("it" being Wunderground). The editing on Wunderground appears to be worse than useless.

Oh, I just checked and AGAIN it has returned the set of values at 11.17 and 11.22 despite my having deleted them on x occasions. Perhaps I should just go in and modify the individual crap items rather than trying to delete the whole line (well, it works so far).

Oh great! Now I can't access my site!!!! 12:10 UTC
Oh, so then the root came back but 500 error on the saratoga folder (php files)
And, now, touch wood, that is back.
Time to get a new host (I keep saying that).