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

20031124: ldm/craft data latencies at NWS Central Region HQ

Karen,

Regarding the LDM computer at the Central Region Headquarters:

    What is its fully-qualified hostname.

    Is there anything unusual in the LDM logfile?

    How many downstream LDM-s are running?

    What is the output from "pqmon".

    Is the data coming-in using LDM-5 or LDM-6 protocols?  You can
    determine this from the LDM's logfile: the downstream LDM will log
    one of the following messages:

        Connected to upstream LDM-5

        Connected to upstream LDM-6

    What LDM is receiving the rtstats(1) information?  Can you send it
    to rtstats.unidata.ucar.edu?

Regarding the LDM computer at the forecast office:

    What is the distribution of latencies (you can estimate this via an
    "ldmadmin watch" on that computer).

    Is there anything unusual in the LDM's logfile?

Are there any firewalls between the forecast office and the Central
Region Headquarters?

What is the bandwidth of the connection from the forecast office to the
Central Region Headquarters?

Is there any chance that I could log into the Central Region
Headquarter's LDM computer as the LDM-user using ssh(1)?

Regards,
Steve Emmerson

> To: address@hidden,
> To: William Gery <address@hidden>,
> To: address@hidden,
> To: Mark Benner <address@hidden>
> From: "Karen Cooper" <address@hidden>
> Subject: ldm/craft data latencies
> Organization: NOAA
> Keywords: LDM CRAFT
>
> We have recently set up a server at the Central Region Headquarters in 
> Kansas City with a number of LDM machines ingesting 88D radar data and 
> sedning them to OU through this server.
> 
> We have noticed very high latencies with this data and have been 
> troubleshooting. All info below if from the server at the headquarters 
> to an individual LDm machine located at a forecast office gathering 
> data.  These numbers are representative of the other machines as well.
> 
> All machines are running RedHat 9 with LDM-6.0.14.
> 
> Standard ping show an average 64ms
> Ping with a packet of 10000 bytes show an average 300ms
> ldmping shows ~.15 elapsed
> traceroute shows 3 hops whithe times of ~.32ms, ~75ms, ~80ms which seems 
> to correspond to the pings fairly well
> ntpdate -q shows a time difference between the machines as ~.03 seconds
> 
> All the LDM machines are running ntpd and the server is running an rdate 
> cron job to a timeserver.
> 
> However, running rtstats on the server we are getting latencies of ~15 
> seconds with some maxs going up to 30 seconds.  This is also reflected 
> by notifyme showing large delays.
> 
> My ldmd.conf file on the server is rather large, with many requests to 
> machines which are not yet online, but paring it down to just a few did 
> not have any impact.  The routers between the machines have the ldm data 
> flagged as high priority, but playing with those settings had no effect, 
> nor does the amount of activity on the connection seem to have any effect.
> 
> I'm hoping you might have some other things for us to try to help 
> troubleshoot this problem.
> 
> 
> 
> -- 
> address@hidden
> Phone:405-366-0434    
> Cell:405-834-8559
> CIMMS/University of Oklahoma
> Affiliate of National Severe Storms Laboratory