[Support #SBB-325304]: Re: 20110712: CONDUIT request -- fire weather grids

Hi Justin,

I just reviewed the real-time stats from the 3 toplevel CONDUIT
machines at the WOC, and two of the machines continue to show
virtually identical volumes of data, and the third shows consistently
less data:

Machines that are virtually identical:

ncep-ldm0.woc.noaa.gov
http://www.unidata.ucar.edu/cgi-bin/rtstats/iddstats_vol_nc?CONDUIT+ncep-ldm0.woc.noaa.gov

ncep-ldm1.woc.noaa.gov
http://www.unidata.ucar.edu/cgi-bin/rtstats/iddstats_vol_nc?CONDUIT+ncep-ldm1.woc.noaa.gov

ncep-ldm2.woc.noaa.gov should show higher data volumes, but shows consistently 
less:

ncep-ldm2.woc.noaa.gov
http://www.unidata.ucar.edu/cgi-bin/rtstats/iddstats_vol_nc?CONDUIT+ncep-ldm2.woc.noaa.gov

Since there was no change in data volumes since you killed the "hung" 
gribinsert process
on ldm2, I am left with the impression that it is not getting the same set of 
"operational"
data that ldm0 and ldm1 are.  Can you check into this again?

Now for some good news:

It appears that the parallel NAM data (i.e., firewx products) data peaks occur 
at
different times than the overall peak in "operational" CONDUIT volume.  This 
means that
the addition of the products is not having as detrimental an effect as they 
would
if the volume peaks were at the same time.  This, in turn, should mean that the 
effect
on the LDM queue residency time does not demand an immediate increase in LDM 
queue size
(which would require an immediate increase in system RAM).

HOWEVER, the fact that the minimum age of the oldest product in the LDM queue 
is 0.5
hours (30 minutes) is worrisome.  We prefer that IDD toplevel, data source, and 
relay
machines maintain a minimum of 1 hour of data in their queues.  It is, 
therefore,
our recommendation that more RAM be added to each of the 3 toplevel CONDUIT 
machines
(at least 2 GB, preferably 4 GB or more) and their LDM queues be increased
correspondingly (i.e., at least 2 GB but preferably 4 GB).  An increase in LDM
queue size will provide for more "elasticity" for the data flow and allow for
more products to be easily added.

Cheers,

Tom
--
****************************************************************************
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: SBB-325304
Department: Support CONDUIT
Priority: Normal
Status: Closed