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

[Support #VXV-940104]: Re: [conduit] missing and incomplete GFS grids on CONDUIT feed

Hi David,

Did you try changing your 5-way split CONDUIT feed requests to 10
as Pete had suggested?

I ask because we have seen a marked improvement in CONDUIT receipt
latencies being reported by freshair2:

http://rtstats.unidata.ucar.edu/cgi-bin/rtstats/iddstats_nc?CONDUIT+freshair2.atmos.washington.edu

We also had occasion to look at the receipt latencies for the NEXRAD2
feed on both freshair1 and freshair2:

freshair1:
http://rtstats.unidata.ucar.edu/cgi-bin/rtstats/iddstats_nc?NEXRAD2+freshair1.atmos.washington.edu

freshair2:
http://rtstats.unidata.ucar.edu/cgi-bin/rtstats/iddstats_nc?NEXRAD2+freshair2.atmos.washington.edu

As you can see from comparing these plots, latencies being reported by freshair1
are significantly less than what is being reported by freshair2.  This is the
same as for the CONDUIT feed.

The questions that come to mind are:

- what is the difference between the network connections to freshair1 and
  freshair2?

  This brings up a couple of questions:

  - is freshair2 using a Gbps Ethernet interface for LDM ingestion?

  - if yes, is the interface actually running at Gbps speed?

    The easiest way to test this is as follows:

    <as 'root'>
    ethtool <ethernet interface>

    We have seen instances where Gbps Ethernet interfaces were actually running
    at 100 Mbps speeds, and/or it was not running full Duplex, and either of 
these
    was greatly impeding the flow of data into the machine.

- how can the latencies on freshair2 be lessened/minimized/mitigated?

  Aside from optimizing some kernel settings related to networking, the
  most straightforward way to mitigate this is to split feed REQUESTs.
  This may help the situation on freshair2 since we see that it is
  requesting all of NEXRAD2 using a single LDM configuration refile
  REQUEST.  I recommend:

  change:

  request NEXRAD2 ".*" idd.unidata.ucar.edu

  to:

  request NEXRAD2 "[02468]/[IES]" idd.unidata.ucar.edu
  request NEXRAD2 "[13579]/[IES]" idd.unidata.ucar.edu

  And then stop and restart your LDM.

Aside:

When looking through the LDM log file on the real server back end
machine of our IDD relay cluster idd.unidata.ucar.edu that is feeding
freshair2, I saw that you are now using a 10-way split REQUEST for
CONDUIT, so my first question above has been answered.



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: VXV-940104
Department: Support CONDUIT
Priority: Normal
Status: Closed
===================
NOTE: All email exchanges with Unidata User Support are recorded in the Unidata 
inquiry tracking system and then made publicly available through the web.  If 
you do not want to have your interactions made available in this way, you must 
let us know in each email you send to us.