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

Re: 20011008: IDD latencies at PSU (cont.)

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 Note that the archive is no longer being updated. We provide the archive for reference; many of the answers presented here remain technically correct, even if somewhat outdated.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: Re: 20011008: IDD latencies at PSU (cont.)
  • Date: Thu, 11 Oct 2001 10:37:31 -0600
"Arthur A. Person" wrote:
> 
> Anne,
> 
> 
> I logged on this evening and started poking around and discovered the
> system had gone into that disk thrashing mode we had worked on several
> months ago but didn't resolve.  I thought maybe this was the whole problem
> altogether, so I stopped the ldm, reset motherlode as feed for WMO stuff
> as well as NMC2 to test the original configuration, rebuild the queue
> (this time with a 600mb queue since the thrashing problem seems to happen
> with larger queues) and then restarted.  It seemed faster to start with on
> data delivery, but again seems to bog down.  So the thrashing thing was
> yet another problem but is not the cause of the current problem unless
> it's somehow related in an underlying way.  Interestingly, I did a little
> diagnosis during the thrashing event and found that by running iostat,
> most of the thrashing was on reads from the disk, something on the order
> of 12,000/sec!  This number seems to be in the 100-300/sec normally.
> 
> I will try switching NMC2 to sunshine for overnight tonight and see how
> things go.  Since the WMO stuff seems to run slow all the time, I don't
> think it will make a difference since NMC2 comes in fast but for only
> certain periods of the day.
> 
> 
> I've forwarded this email to Jeff Wolfe for comments.  When complete, I'll
> forward them to you.
> 
>                                   Art

Art,

I guess at this point then we're waiting to hear back from Jeff.

Regarding a large queue, how long were you running with the 1G queue
before this?  And, have you ever tried or are you interested in trying
to build a large queue using the large file support?  I believe Daryl
Herzmann at UIowa is running a 3-4Gb queue on his Redhat Linux 7.1
machine that serves as a short term archive, and has been quite
pleased.  Just a FYI.

Anne 
-- 
***************************************************
Anne Wilson                     UCAR Unidata Program            
address@hidden                 P.O. Box 3000
                                  Boulder, CO  80307
----------------------------------------------------
Unidata WWW server       http://www.unidata.ucar.edu/
****************************************************