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

Re: 20011008: IDD latencies at PSU (cont.)

This archive contains answers to questions sent to Unidata support through mid-2025. Note that the archive is no longer being updated. We provide the archive for reference; many of the answers presented here remain technically correct, even if somewhat outdated. For the most up-to-date information on the use of NSF Unidata software and data services, please consult the Software Documentation first.

  • Subject: Re: 20011008: IDD latencies at PSU (cont.)
  • Date: Wed, 17 Oct 2001 10:29:09 -0600
Hi Jeff,

Jeff Wolfe wrote:
> 
> >
> > I began to wonder if the number of products is part of the problem.  The
> > number of products site A can transfer to site B is fundamentally
> > constrained by the number of rpc calls that A can make to B.  Here's a
> > piece of an early email from Russ regarding this subject:
> 
> Given the nature of the problem and corresponding packet loss on our end,
> it's quite possible rpc is being affected. If a product is contained in one
> rpc call, and it's a larger product, it's more likely that one of the TCP
> packets associated with the rpc call will be lost, forcing a retransmit.
> 

Bear in the mind that the ldm sends products in blocks of 16K or less. 
Small products are sent in their entirety, while large products are
broken up.  Still, as I understand it, if a portion of a large product
is lost, the entire product is resent.

One way to see if this is happening is to turn on verbose logging to the
appropriate rcp.ldmd process, in this case it's the rpc.ldmd that is
feeding from motherlode.  This will fill up the logs quickly, but you
can see the pieces as they arrive.  To cycle through the
quiet->verbose->debug log modes, send the appropriate rpc.ldmd process a
USR2 signal via 'kill'.



> 
> >
> > I thought I'd try the same thing to nutone for comparison, but I still
> > can't connect there:
> >
> > motherlode.ucar.edu% ldmping -i5 nutone.ems.psu.edu
> > Oct 16 16:59:22      State    Elapsed Port   Remote_Host
> > rpc_stat
> > Oct 16 16:59:35      NAMED  12.223594    0   nutone.ems.psu.edu  can't
> > contact portmapper: RPC: Timed out
> > Oct 16 16:59:52      NAMED  12.040442    0   nutone.ems.psu.edu  can't
> > contact portmapper: RPC: Timed out
> > Oct 16 17:00:09      NAMED  12.258855    0   nutone.ems.psu.edu  can't
> > contact portmapper: RPC: Timed out
> > Oct 16 17:00:26      NAMED  12.419979    0   nutone.ems.psu.edu  can't
> > contact portmapper: RPC: Timed out
> 
> Sorry, I think this is fixed now.. (I hope it is.) Can you check and let me
> know?
> 

Yes, much better!  Thanks!

> 
> By the way, I can create an account on nutone for you guys, if you want.
> 
> -JEff


Thank you for the offer.  I'll consider it.

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