Closed uandrae closed 3 years ago
Can you narrow it down to the vobs files that lead to the problem and attach them to this issue?
It can be reproduced by the added file
Wed Aug 26 13:02:55 GMT 2020
load proj4 5.2.0 (PATH, MANPATH, PROJ4_DIR, EC_FFLAGS, EC_CFLAGS, EC_CXXFLAGS, EC_FLDFLAGS,EC_CLDFLAGS, EC_CXXLDFLAGS)
load R 3.6.3 (PATH)
Reading: /scratch/ms/se/snh/vobs/prod_SPP_defpert_43h2_1//vobs2019060100
***
New SQLITE obs file created: /scratch/ms/se/snh/harp/OBSTABLE/OBSTABLE_2019.sqlite
***
Writing to: SYNOP in /scratch/ms/se/snh/harp/OBSTABLE/OBSTABLE_2019.sqlite
Writing to: TEMP in /scratch/ms/se/snh/harp/OBSTABLE/OBSTABLE_2019.sqlite
[1] "FAILURE dbSendStatement: 1 / 20"
[1] "no such column: NA"
[1] "FAILURE dbSendStatement: 2 / 20"
[1] "no such column: NA"
[1] "FAILURE dbSendStatement: 3 / 20"
Fixed by #35
Problem was caused by pressure levels set to -99 in the TEMP section of the vobs file. Apparently it hasn't come across that issue before!
Ah, interesting. It must be a feature of my new EcCodes based BUFR reading. I assume it's not to tricky to fix. In any case I have to check the behavior compared to the old code on my side.
The following run on ecgate stops after 3days. The sqlite is readable, but incomplete.
Generates