Has anybody noticed some rather odd filenames created from
data ingested by the ldm just before 00Z Dec 1 today? Here's an
example of METAR reports that are FILEd by the LDM. Size in
bytes, date and filename:
852802 Dec 1 21:24 metar.20001201.1900.asc
560370 Dec 1 21:24 metar.20001201.2000.asc
374984 Dec 1 21:24 metar.20001201.2100.asc
112658 Nov 30 23:59 metar.20040401.0000.asc !!
I also got the same odd filename (date of 04/01/2004) with
syntopic data being filed with a similar pattern/action at
about the same time - just before 00Z Dec 1.
And it's not a complete fluke considering there is a fair
amount of data in the oddly-named file. A lot of times you'll
get some bad filenames because the day is wrong in the bulletin
but the year and month are never this far off. All the bulletins
that were filed in the odd file had the DDHHMM field of 010000
(SAUS80 KWBC 010000).
Here is the entry from the pqact.conf file:
WMO ^S[AP].* .... ([0-3][0-9])([0-2][0-9])([0-5][0-9]) FILE
raw_data/obs/metar/(\1:yyyy)(\1:mm)\1/metar.(\1:yyyy)(\1:mm)\1.\200.asc
Another part that is very strange is that these files are in the
"correct" directory of 20001201/ which is created using the same
information from the WMO header. In other words, the same
pattern was used to make the directory and filename but the LDM came
up with different values for each:
(\1:yyyy)(\1:mm)\1/metar.(\1:yyyy)(\1:mm)\1 ---> 20001201/metar.20040401
I saw the same behaviour on 4 different systems, all which
use the same or similar pattern/action, and the same version
of the ldm - 5.0.9 (I know, I'm going to upgrade soon :)
-jim
--
Jim Cowie
WITI/Lifeminders, Inc.
303-497-8584 (office)
720-231-7948 (cell)
