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

[LDM #YPJ-917456]: Slow LDM delivery problem



Brice,

Tom's out of the office for most of this week.

> Our SMG customers have a continuing problem in getting products from
> our LDM distribution server to their NWS LDAD.  It may take upwards of
> 30 minutes to an hour to get some products transferred.  They started
> noticing this right after an upgrade to their LDAD and its outward facing
> firewall/boundary devices.  However, direct file transfers don't show
> as great a lag as transfers in the LDM stream.

Did the problem exist before the LDAD upgrade?

What version of the LDM does the LDAD system run?

What version of the LDM are you running?

> To further complicate things, file transfers into the distribution server
> from other nodes on our LAN, all with Gig-E interfaces, seem slow as well.
> And slower at times than others, e.g. a 8K file may take 0 seconds to
> transfer at times or as much as 20 seconds.  And all this takes place
> with a high idle time on the workstation and no major load factor showing.

Can you monitor the network flow?  Is there congestion?

> We are running a 2GB queue and have 17 feeds from the server, 3 of them
> from boxes in our network.  The server has 12 LDM requests, 5 of them
> from internal sources.

Your LDM server has 12 incoming-data connections and 17 outgoing-data 
connections.  Is this correct?

> I have included some simplified drawings of the network connections with
> transfer speeds.
> 
> I am wondering if there is a problem with the number of feeds interacting
> with the number of products that flow through the queue.  We are flowing
> most of the NOAAPort stream through it, with the exception of the GRIB2
> data,  along with additional surface obs from AFWA.  The SMG is also
> putting selected GRIB2 and BUFR product into the queue from their local
> modeling cluster.  So, a combination of many small products and some
> larger ones - the GRIB files.  We have had to adjust the number of slots
> on the various LDM queues in play in the system to account for the high
> number of small messages.

What's the maximum latency parameter given to your LDM server?  Are you using 
the default one hour?

What does periodic execution of the pqmon(1) utility show for the age of the 
oldest product in the queue?

> Thoughts?  Suggestions?
> 
> Thanks in advance,
> 
> Brice
> 
> Brice Biggerstaff
> MIDDS Software Support
> 281-853-3011 (w)
> 713-764-2601 (p)
> address@hidden  (alpha pager for text and email)
> 
> Res Confacti Erimus
> "We Get Things Done!"

Regards,
Steve Emmerson

Ticket Details
===================
Ticket ID: YPJ-917456
Department: Support LDM
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.