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

Re: ldm dropping old data

"Fingerhut, William A" wrote:
> 
> Anne,
> 
> I have finally returned, unpacked, and gotten back to work.
> The workshop was really good; I learned a lot.
> 
> When we talked about us (LSC) loosing data over 1 hour old,
> I thought you said that I could use the -m flag on the
> rpc.ldmd invocation. When I read the man page for rpc.ldmd
> I wasn't so sure. 'skipping ahead in the queue' seems
> strange? I would feel better if I understood the man page
> remarks? Any chance you could help me out?
> 
> Bill
> 
> ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
> < Bill Fingerhut, Professor           PHONE: 802-626-6257 >
> < Meteorology Dept                    FAX:   802-626-9770 >
> < Lyndon State College                                    >
> < Lyndonville, Vt 05851                                   >
> <                                                         >
> < EMAIL:     address@hidden                  >
> <            address@hidden                  >
> < WWW:       http://apollo.lsc.vsc.edu/                   >
> <                                                         >
> < disclaimer: I know nothing - I only work here.          >
> vvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvv

Hi Bill,

Sorry for the delay in responding - I just returned from my vacation. 
And, thanks for the kind words about the workshop.  It's great to hear
that.  I'm glad you found it helpful.  Did you enjoy the rest of your
vacation?

The man page is confusing about the -m and the -o option to rpc.ldmd.  I
think the "skipping ahead in the queue" phrase refers to the case where
an upstream site sends a downstream site a product that is older than
the downstream site wants.  In that case, the downstream site will send
the upstream site a RECLASS message which will cause the upstream site
to "jump forward in the queue", i.e., send the downstream site newer
products because the older products are too old. 

In general, the -m flag specifies how old of a product you are willing
to accept.  I have used it to request products well over one hour old. 
One wrinkle is if your upstream site doesn't have products as old as you
are requesting.  In that case you will only get what your upstream site
has.  Another possibility to consider if you're going to request very
old products is that your site may get swamped handling the old data and
not be able to keep up locally or relay more timely data promptly.  Are
you a leaf node?  From our site contact list it appears that Universite
du Quebec a Montreal is feeding from you.  Is that still accurate?

In your case you probably only need to use the -m option.  Just for
completeness, the -o option can be used in conjunction with the -m
option (but not the other way around) to limit requests to even more
recent data.  This is why the man page says "The  value  of  toffset 
should  be strickly [sic] less than max_latency."  The only situation I
can think of where this is useful is the one mentioned in the man page:
"This may be useful when bootstraping an empty  queue  and  you  are 
not interested in last hour's  data."  

I hope I've been able to shed some light on this for you.  Please let me
know if you have further questions.

Anne

-- 
***************************************************
Anne Wilson                     UCAR Unidata Program            
address@hidden                 P.O. Box 3000
                                  Boulder, CO  80307
----------------------------------------------------
Unidata WWW server       http://www.unidata.ucar.edu/
****************************************************