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

Re: 19990923: AVN model output



Wayne,

Sure, if you get a subset of data then the downstream sites will get a
subset of data. But looking at the topology/statistic page they are
already loosing data because of large latencies. ie

ocs.oce.orst.edu 52.3 Minutes
        chinook.phsx.ukans.edu 52.0 Minutes
           quasar.phsx.ukans.edu unknown

http://www.unidata.ucar.edu/projects/idd/status/idd/fosTopo.html


The 2 request lines will not get a subset of data.

Robb...

On Thu, 23 Sep 1999 address@hidden wrote:

> Rob,
> 
> Forgive my ignorance, but won't that affect those sites downstream that are 
> feeding off of me?
> 
> Wayne
> 
> 
> > From address@hidden Thu Sep 23 14:17 PDT 1999
> > X-Authentication-Warning: wcfields.unidata.ucar.edu: rkambic owned process 
> > doing -bs
> > Date: Thu, 23 Sep 1999 15:17:15 -0600 (MDT)
> > From: Robb Kambic <address@hidden>
> > To: Steve Chiswell <address@hidden>
> > cc: address@hidden, address@hidden
> > Subject: Re: 19990923: AVN model output
> > MIME-Version: 1.0
> > 
> > Wayne,
> > 
> > Suggestions are  to only request the data that you really need or
> > to split you request lines up to the same feeder, one using the the
> > hostname and the other using the IP number. You can also look for other
> > feeder sites but lately they are hard to come apon.  The web page has the
> > FOS topology:
> > 
> > http://www.unidata.ucar.edu/projects/idd/status/idd/fosTopo.html
> > 
> > 
> > Most of the top and first tier sites are already at their capacity for
> > downstream feeders. 
> > 
> > Robb...
> > 
> > On Thu, 23 Sep 1999, Steve Chiswell wrote:
> > 
> > > Wayne,
> > > 
> > > Looking at the stats files from your computer (you'll find these in
> > > ~ldm/logs/*.stats) which the mailpqstats program sends here,
> > > I see that your latencies have been exceeding 1 hour for the HDS feed-
> > > which should be accompanied by RECLASS statements in your ldmd.log file.
> > > 
> > > The RECLASS statements mean that your queue has fallen over 1 hour behind,
> > > and so has jumped ahead in hopes of trying to catch up to present. As
> > > a result, you will lose data when those jumps occur.
> > > 
> > > The data should already be purged out of Washington's data queue since it
> > > is past several hours old- so they will no longer have it in their data
> > > queue.
> > > 
> > > I have posted todays AVN thinned grib files at:
> > > ftp://ftp.unidata.ucar.edu/pub/chiz/99092300_AVN.wmo  & 99092312_AVN.wmo
> > > 
> > > I will CC Robb Kambic on this message to see if he has any suggestions for
> > > you
> > > as far as improving your data latency.
> > > 
> > > Steve Chiswell
> > > Unidata User Support
> > > 
> > > On Thu, 23 Sep 1999 address@hidden wrote:
> > > 
> > > > Chiz,
> > > > 
> > > > Whenever I modify the pqact.conf file, I run the check "pqact".  Here 
> > > > is a run
> > > > and the results
> > > > 
> > > > 
> > > > ocs 148: pqact -vxl- -q /dev/null pqact.conf
> > > > Sep 23 19:16:31 pqact[19178]: Starting Up
> > > > Sep 23 19:16:31 pqact[19178]: Successfully read pattern file 
> > > > "pqact.conf"
> > > >         Mapping 16384
> > > > Sep 23 19:16:31 pqact[19178]: mmap: 0 0 16384: Resource temporarily 
> > > > unavailable
> > > > Sep 23 19:16:31 pqact[19178]: pq_open failed: /dev/null: Resource 
> > > > temporarily unavailable
> > > > Sep 23 19:16:31 pqact[19178]: Exiting
> > > > 
> > > > So there were no errors found.  Here is the entriy from pqact.conf
> > > > 
> > > > HDS     ^HH[KL].50 KWB. (..)(..).*(/mAVN|/mSSIAVN)
> > > >         FILE    -close data/GRIB/(\1:yy)(\1:mm)\1\2_avn-hgt.39-40
> > > > 
> > > > I just checked again for the most recent output and it is
> > > > 
> > > > -rw-r--r--   1 ldm         18040 Sep 23 10:37 99092312_avn-hgt.39-40
> > > > 
> > > > Dumping the headers reveals
> > > > 
> > > > gribdump -h 99092312_avn-hgt.39-40 | egrep '( Grid|Time 2)'
> > > >                        Grid : 39
> > > >                 Time 2 (P2) : 12
> > > >                        Grid : 40
> > > >                 Time 2 (P2) : 12
> > > >                        Grid : 40
> > > >                 Time 2 (P2) : 42
> > > > 
> > > > So for grid 39, I receved hour 12, and for grid 40 received hours 12 
> > > > and 42.  That
> > > > was about 2 hours ago so I can assume there's nothing else coming down 
> > > > the pipe.
> > > > 
> > > > I contacted Harry Edmon and he indicated that the product was received 
> > > > at U of Washington.
> > > > Is there a way that I can re-request that product via ldm from my 
> > > > upstream host
> > > > grayskies.atmos.washington.edu and see if that works?  Can you think of 
> > > > any reason why I
> > > > wouldn't have received it in the 1st place?
> > > > 
> > > > Thanks,
> > > > 
> > > > Wayne Gibson
> > > > Oregon State University
> > > > 
> > > 
> > > 
> > 
> > ===============================================================================
> > Robb Kambic                            Unidata Program Center
> > Software Engineer III                          Univ. Corp for Atmospheric 
> > Research
> > address@hidden                 WWW: http://www.unidata.ucar.edu/
> > ===============================================================================
> > 
> > 
> 

===============================================================================
Robb Kambic                                Unidata Program Center
Software Engineer III                      Univ. Corp for Atmospheric Research
address@hidden             WWW: http://www.unidata.ucar.edu/
===============================================================================