[LDM #PFK-426667]: Load average goes above 10 on weather3.admin.niu.edu running LDM

Hi Gilbert,

re: high iowait the root of the problem
> No, the root of the problem...
> 
> /home/ldm% w
> 08:46:07 up 20:33,  3 users,  load average: 1.00, 1.47, 1.50
> USER     TTY      FROM              LOGIN@   IDLE   JCPU   PCPU WHAT
> ldm      :0       -                Wed12   ?xdm?   1:57m  0.06s -/bin/tcsh -c 
> /
> ldm      pts/1    :0.0             Wed12   16:01m  0.47s  0.28s -tcsh
> ldm      pts/2    weather.admin.ni 08:46    0.00s  0.07s  0.00s w
> 
> Was this: For some reason, the queue was 2.1 GB, even though I had
> specified 1.2 GB in the registry! Going into the registry, dropping it
> down to 1.2 GB forcibly, and reducing the number of blocks by 90%,
> stopped the issue from happening.
> 
> How did that happen?

It happened because you had a registry.xml configuration set
to 'decrease maximum latency'.  This caused the LDM to automatically
resize the queue to increase the queue residency time.  I just
edited your registry.xml file and set the configuration to what
I always use 'do nothing'.  This will stop the LDM from resizing
your queue in the future.  Given the low load averages now, I
think that keeping your queue to a fixed size, manageable size
is what is called for.

Musing: the problem may have appeared because you keep your
LDM queue in shared memory...

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: PFK-426667
Department: Support LDM
Priority: Normal
Status: Closed