NOTICE: This version of the NSF Unidata web site (archive.unidata.ucar.edu) is no longer being updated.
Current content can be found at unidata.ucar.edu.

To learn about what's going on, see About the Archive Site.


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

[LDM #QYR-744357]: increase Q size



Virginia, Chris, Robert,

> It shows up as "latency" on the NWS website, however, I believe the real
> problem is duplicate data being re-downloaded as the TGP48/49 server (w/
> 2GB queue) switches primary/alternate between the two sources (RDS and TDS
> w/ 4GB queues). LDM seems to start up a new connection when switching
> sources and data that has aged out of the smaller downstream queue is
> re-downloaded from the larger upstream queue. The duplicate data appears as
> a false latency. Please correct me if I'm wrong as most of my LDM knowledge
> if from trial and error....mostly error.

I suspect the large "latencies" you're seeing are due to the size of the 
product-queue being incompatible with the maximum allowed latency parameter, 
which is 1 hour by default. If the product-queue is too small to hold more than 
30 minutes worth of data, then a duplicate data-product that arrives 45 minutes 
late will be accepted as a new data-product because its previous arrival has 
been purged from the product-queue.

Beginning with LDM version 6.9, the LDM system can now automatically reconcile 
the size of the product-queue and the maximum latency parameter, either by 
increasing the size of the queue, decreasing the maximum latency parameter, or 
notifying the LDM user via email -- depending on the value of the 
"reconciliation-mode" parameter. You should be able to find a lot of 
support-email on this subject <http://www.unidata.ucar.edu/software/ldm/#help> 
by searching for "reconciliation-mode". Also, this feature is documented in the 
online LDM reference 
<http://www.unidata.ucar.edu/software/ldm/ldm-6.10.1/basics/index.html>.

> I know there is a flag in LDM to limit the downloading of data to that
> below a certain latency threshold, however, you risk missing data from a
> site that is experiencing actual latency (I've seen individual sites with
> real latency over 3000 seconds due to telco problems).
> 
> --
> Chris Calvert
> Software Engineer
> National Weather Service
> WSR-88D Radar Operations Center
> 1313 Halley Circle, Norman, OK 73069
> (405) 573-3323

Regards,
Steve Emmerson

Ticket Details
===================
Ticket ID: QYR-744357
Department: Support LDM
Priority: Normal
Status: Closed