[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Support #TOL-308327]: statistics

Hi Yoshihiro,

> I just changed the LDM queue to 2G. I will be watching the
> results.

Very good, thanks!

> Good news and results --> Even before the queue size
> change (but correcting the mistype error), the 12:00 Data
> files  was received completetly fine and now the MM5 and
> WRF models are running without any problem.

This is good news.  Even though things were running smoothly, there
was still a problem with the queue being too small.  The problem
can be seen in the latency plots for two periods on Saturday, June
11.  The envelope of higher latency products seen in these two
periods indicated that some products were being sent to atm77
even though they had been received earlier.  The reason for the
resend was the queue was not big enough to hold the data for
a period long enough for the duplicate product detection and rejection
to work.

> I would appreciated if you could watch the statistics and
> see if seems to be fine now.

Everything looked good until the transmission of the 12 UTC run today.
For some reason -- and we do not know why -- the latencies in the data
being sent climbed to over 4000 seconds.  It could be the case that
network traffic at your university increased to the point that getting the
data became problematic.  It is also possible that there was a problem
in getting data out of the cise-nsf.gov domain.  In order to figure out
which possible reason for the increased latencies is the correct one,
we need to you to reconfigure your ~ldm/etc/ldmd.conf request lines
again, and we need you to load the latest version of the LDM, LDM-

The latest version of the LDM can be grabbed using anonymous FTP
from our FTP server:

machine:   ftp.unidata.ucar.edu
user:      anonymous
pass:      address@hidden
directory: pub/ldm/beta
file:      ldm-

Please build and install LDM- in the normal manner:

<as 'ldm'>
cd ~ldm
tar xvzf ldm-
cd ldm-
make install
sudo make install_setuids
cd ~ldm
ldmadmin stop
rm runtime
ln -s ldm- runtime

Before restarting the LDM (or before stopping the running of your
existing LDM), please make the following changes in your ~ldm/etc/ldmd.conf

- replicate your 5 CONDUIT request lines to idd.cise-nsf.gov and change
  the upstream host name to idd.unidata.ucar.edu

After making this change, you will have 10 request lines for the CONDUIT
data.  You do not need to worry about specifying PRIMARY or ALTERNATE
on any of the request lines when running LDM-6.4.x since this LDM
version does autoswitching of the feed mode for each duplicate connection
in order to make PRIMARY the feed that is delivering data the fastest.
Given the autoswitching strategy, you will not be using twice the
bandwidth for one set of connections.  The nice thing about this feature
is a site can redundantly request the same data from more than one upstream
host, and then not have to worry if any one host becomes unavailable.

Please let us know if you have any questions on what we are asking you to do.


Unidata User Support                                    UCAR Unidata Program
(303) 497-8642                                                 P.O. Box 3000
address@hidden                                   Boulder, CO 80307
Unidata HomePage                       http://www.unidata.ucar.edu

Ticket Details
Ticket ID: TOL-308327
Department: Support IDD
Priority: Normal
Status: Closed