meteohub log

[ralf2201]

Hi,

I use a RFXCOM and the meteohub run on a alix1d. and have a lot of errors in this logfile. I just migrated from a NSLU2 and thought there was a problem with the usb stick. But no the probleme is still there. Can anybody help me on this?

logger (21.02.2012 23:41:37): station 0 (RFXCOM), wrong checksum (34 vs computed 24) for sensor model 2a19 (PCR800) in byte sequence: 5c 2a 19 04 45 00 00 40 70 00 40 63 03
loggereval (22.02.2012 04:47:07): illegal data record in line 1344 while reading from file /data/weather/200911/raw: 0091101033332 wind0 292 0 0 0043
loggereval (22.02.2012 04:47:07): illegal data record in line 1352 while reading from file /data/weather/200911/raw: 2009110033502 rain0 0 0 1879
loggereval (22.02.2012 04:47:07): illegal data record in line 4509 while reading from file /data/weather/200912/raw: :00912011159

[ralf2201]

Hi,

i saw just now that there are errors in my raw data. Could this cause the problem and is there a way to delete or modify this errors.


0091227195504 wind0 180 36 28 0005
20091227195505 th0 37 68 -017
20091227195507 rain0 0 0 3416
2009122795518 wind0 225 39 28 0003
20091227195532 wind0 "02 24 28 0014
20091227195546 wind0 180 36 28 0005
2091227195554 rain0 0 0 3416
20091227195600 th0 37 68 -017
20091227195600 wind0 18 29 34 000
20091227195614 ind0 180 45 34 0000
20091227195628 wind0 225 41 34 0002
20091227195641 rain0 0 0 3<16
20091227195642 wind8 202 35 34 0006
20091227195653 th0 37 67 -019
20091227195656 wind0 270 26 30 0013
20091227195710 wind0 222 32 30 0008
20091227195724 wind0 158 22 30 0016
20091227195728 rain0 0 0 3416*20095227195738 wind0 248 17 30 0022
20091227195746(th0 37 67 -019
20091227195752 wind0 180 17 21 0022

[YJB]

I've gone through a similar situation in the past.

The official way of doing things is to use the gui and mark the offending records for deletion. Personally I find this process very tedious and inefficient.

I've used a text editor (like ultraedit) and modified the files manually (you could use vi as well of course). Once you are finished, just run a re-computation and all your aggregates will build in a sane way. As long as you are not touching the current month -- and -- you are not issuing a re computation, you can do the modification without taking MH offline

Disclaimer
he above is, of course, not supported by Boris, so you are basically on your own. A backup of the original data-files is highly recommended.

[ralf2201]

Anybody an idea what can cause this errors. Is it possible that it was a problem of a damaged filesystem with the usb stick?

[YJB]

Yes, most likely it was an issue with the storage, but I've also seen it happening during hiccups of MH in the past.

From what you've posted, it seems that this corruption happened back in 2009, if you do not see it in more recent datasets, I would just go ahead, fix it and move on.

If the corruption also occurs in more recent versions of the data, I would be indeed concerned.