Hello all,
For those of you who used the EMWIN feed under the EXP header on
www.ehs.niu.edu...
I had to rename the machine to www2.ehs.niu.edu. For those of you feeding
EMWIN from the previous machine, www.ehs.niu.edu, please make a note of
it. Thanks!
In the process, I upgraded that machine, and all my other ones ingesting
NOAAport (weatherx.admin.niu.edu), including my satellite receiver
machine, to ldm-6.7.0.3. Although it is still a beta version, I have
noticed a roughly 10%-33% improvement on average, and a lot more under
heavy load, to the GEMPAK dcgrib2 decoder. During peak (or even high)
periods of incoming model data, the dcgrib2 decoder would max out a CPU
on one of my machines. Now, it doesn't even come close. As a result, the
load average is considerably lower during high volume model ingest times,
and obviously, I like that. This also improves the load on the McIDAS
decoders as well, though nowhere near as dramatic as dcgrib2 from GEMPAK.
Otherwise, as has been the case with ldm-6.7.0.1, this version appears to
be very stable, and I like the improved ldmd.conf and pqact.conf error
detection, and the other improvements made to the log file messages.
However, I am mildly amused that the UNIDATA stats page:
http://www.unidata.ucar.edu/cgi-bin/rtstats/siteindex?www2.ehs.niu.edu
shows that I am running LDM 6.4.5 on www2.ehs.niu.edu, as well as getting
the CONDUIT feed. Rest assured neither are true, so I don't know what's
going on there. Otherwise, all my other machines show the correct
version/data streams on the rtstats page.
*******************************************************************************
Gilbert Sebenste ********
(My opinions only!) ******
Staff Meteorologist, Northern Illinois University ****
E-mail: sebenste@xxxxxxxxxxxxxxxxxxxxx ***
web: http://weather.admin.niu.edu **
*******************************************************************************