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

[IDD #QBL-783661]: split LDM REQUEST for CONDUIT feed

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 Note that the archive is no longer being updated. We provide the archive for reference; many of the answers presented here remain technically correct, even if somewhat outdated.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: [IDD #QBL-783661]: split LDM REQUEST for CONDUIT feed
  • Date: Mon, 16 Sep 2019 21:36:22 -0600
Hi Rick,

This is a follow-up to the email that Dave Parker sent
you earlier this evening...

I _think_ that the 5-way split for the CONDUIT feed that
I typed in the LDM configuration file on npserv1 was missing
a crucial piece: a '$' (dollar sign) at the end of the extended
regular expression.

The REQUESTs should read:

REQUEST CONDUIT "[05]$" idd.unidata.ucar.edu
REQUEST CONDUIT "[16]$" idd.unidata.ucar.edu
REQUEST CONDUIT "[27]$" idd.unidata.ucar.edu
REQUEST CONDUIT "[38]$" idd.unidata.ucar.edu
REQUEST CONDUIT "[49]$" idd.unidata.ucar.edu

The other thing I should have added is the need/advice to
check the latencies being reported by the upstream feed
hosts that you are REQUESTing feeds from to make sure that
the latencies shown in plots for your machines are not,
in fact, seen further up the data path.  A REQUEST to
our idd.unidata.ucar.edu cluster will actually result in
your machine being fed from one of the real-server backends
of the cluster.  The real-server backend machines for
idd.unidata.ucar.edu all have names that follow:

node<n>.unidata.ucar.edu

examples:

node0.unidata.ucar.edu
node1.unidata.ucar.edu
 ...
node7.unidata.ucar.edu

Since these machines have the identical setup and are fed
in exactly the same way, their latencies should be virtually
identical, so you can look at any of them to get an idea
of what the upstream latencies are.  For instance:

http://rtstats.unidata.ucar.edu/cgi-bin/rtstats/iddstats_nc?CONDUIT+node3.unidata.ucar.edu

I can explain why splitting feeds results in lower latencies
at the MUG meeting tomorrow if you are interested.

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
****************************************************************************