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

Re: 19991115: IDD Slowdowns

Tom,

This is a hard question to answer for certain, but he's a suggestion. When
the latencies are high, check the latency at Cornell.  The topo/latencies
are at :

http://www.unidata.ucar.edu/projects/idd/status/idd/fosTopo.html

Currently:

snow.cit.cornell.edu 6.0 Minutes
       ldmhost.dri.edu unknown
          satellite.usask.ca unknown
              tornado.usask.ca unknown
          monsoon.colorado.edu 0.1 Minutes
          coldwave.dri.edu unknown
          sift.dri.edu unknown
          climate.dri.edu unknown
       shear.cit.cornell.edu unknown
       vortex.weather.brockport.edu 8.6 Minutes

Brockport is only 2.6 minutes behind.

You can also do an ldmping to Cornell at different time during the day.
If you discover that it's the network, contact your network folks or ISP
when the latencies are the highest.  On the otherhand, if you are only a
couple minutes behind Cornell then further investigation into the latency
tree or thelma's ingestion will be necessary. Currently the NOAAport
ingestion is 1-2 minutes behind, thelma is about 45 minutes behind.

Robb...

On Mon, 15 Nov 1999, Unidata Support wrote:

> >To: address@hidden
> >From: Tom McDermott <address@hidden>
> >Subject: IDD Slowdowns
> >Organization: .
> >Keywords: 199911152254.PAA00211
> 
> Hi IDD Support,
> 
> Could the slowdowns we have experiencing at certain times of the day
> over the last several months be related to the problem discussed in these 
> messages, or would this just be internal to NWS?  I realize there are
> other factors involved: the fire on the Cray 3, problems at thelma,
> latency generally very poor in the Northeast, etc.  Thanks.
> 
> Tom McDermott
> ------------------------------------------------------------------------------
> Tom McDermott                         Email: address@hidden
> System Administrator                  Phone: (716) 395-5718
> Earth Sciences Dept.                  Fax: (716) 395-2416
> SUNY College at Brockport
> 
>  
> ---------- Forwarded message ----------
> Date: Sat, 13 Nov 1999 14:51:44 -0500 (EST)
> From: Local Data Manager <ldm>
> To: rochette, tmcderm
> Subject: NWS Admin Message
> 
> 
> 505 
> NOUS72 KNCF 131912
> ADMNCF
> DUE TO THE EXCESSIVELY HIGH VOLUME OF PRODUCTS ON THE NOAA
> WEATHER WIRE, GTE HAS SHUTDOWN THE WEATHER WIRE UPLINKS AT 
> WFO KANSAS CITY (EAX) AND WFO TULSA (TSA).
> EFFECTIVE IMMEDIATELY, ALL OFFICES MUST CUT BACK ON THE
> NUMBER OF SWR PRODUCTS ISSUED.
> ACCORDING TO GTE, THE NWW UPLINKS ARE BEING OVERWHELMED BY
> THE VOLUME OF PRODUCTS BEING UPLINKED.  THE BIGGEST OFFENDER
> IS THE SWR PRODUCT WHICH IS UPLINKED 2 AND SOMETIMES 3 TIMES
> AN HOUR BY EACH SITE.
> THE SWR PRODUCT IS CREATED VIA AN FXA CRON ON THE DS1 SERVER
> BY THE /AWIPS/ADAPT/ADAPT_APPS/BIN/HWRNWWS PROGRAM.
> AS DELIVERED IN THE AWIPS SOFTWARE, THE HWRNWWS PROGRAM WAS
> SCHEDULED TO RUN ONCE PER HOUR AT 10 MINUTES AFTER THE HOUR.
> MOST SITES HAVE MODIFIED THE FXA CRON ON DS1 SUCH THAT THE 
> HWRNWWS PROGRAM IS RUN 2 OR MORE TIMES PER HOUR.
> UNFORTUNATELY, THE CUMULATIVE EFFECT OF ALL AWIPS SITES 
> RUNNING THE HWRNWWS PROGRAM MORE THAN ONCE PER HOUR IS
> OVERWHELMING THE NOAA WEATHER WIRE UPLINKS.
> TO DETERMINE IF THE HWRNWWS PROGRAM IS RUNNING MORE THAN
> ONCE AN HOUR, USE THE FOLLOWING PROCEDURE:
> 1.  log into ds1
> 2.  cd /data/logs/fxa/991113
> 3.  ll hwrnwws*
> IF THERE'S MORE THAN ONE FILE PER HOUR, THE FXA CRON NEEDS TO BE 
> CHANGED.  IF THERE'S SOMEONE AT YOUR OFFICE WHO CAN MODIFY THE
> DS1 FXA CRON SUCH THAT THE HWRNWWS PROGRAM IS RUN ONCE PER HOUR,
> PLEASE MAKE THE CHANGE.  OTHERWISE CONTACT THE NCF TO MAKE THE 
> CHANGE TO THE FXA CRON.
> 
> ----------------------------------------------------------------------------
> 
> Date: Mon, 15 Nov 1999 17:22:38 -0500 (EST)
> From: Local Data Manager <ldm>
> To: rochette, tmcderm
> Subject: NWS Admin Message
> 
> 
> 166 
> NOUS72 KNCF 152204
> ADMNCF
> NWSH is aggressively pursuing a solution to the current NWWS problem.
> In gereral this problem appears to be related to traffic volumes at
> peak periods of the hour reaching too high a level.
> This is contributed to by the generation of multiple copies of some
> products at some sites.
> We are woking to implement a fix to this problem as soon as possible.
> We also ask sites minimize any multiple NWWS products they may be
> generating.
> We will inform the field as soon as a fix has been implemented.
> Fred Branski
> 

===============================================================================
Robb Kambic                                Unidata Program Center
Software Engineer III                      Univ. Corp for Atmospheric Research
address@hidden             WWW: http://www.unidata.ucar.edu/
===============================================================================