Hi Tyler,
re:
>Since my system looks to be mimic'ing Jeff's I can chime in and say that
>it is definitely not IO related with my L3 system.
>
>My queue file is running in a memory based file system with the entire
>queue loaded into memory via a tmpfs file system. That completely
>eliminates the IO issue you would see with placing the queue on disk.
>
>The only IO hit I take is when taking the data out of queue and writing to
>disk which is dedicated to LDM data. I have no downstream feeds.
This information eliminates I/O bottleneck concerns from your machine,
at least.
>pqmon shows max age of 6902 which leads me to believe the queue size is
>plenty big.
Yup, I agree.
>There has got to be an intermittent upstream issue somewhere. Those are
>always the worst ones to troubleshoot!
Given the latencies seen (excluding round-trip or recirculated
products), I would imagine that the problem is at the source -- when
the products are broadcast in NOAAPort. This view is buoyed by an ADM
message that Darryl Herzmann sent along a little bit ago:
NOUS72 KNCF 111452
ADMNCF
THE NWSTG IS EXPERIENCING TECHNICAL DIFFICULTIES WITH A FRONT END
PROCESSOR. THEY ARE TROUBLSHOOTING THE ISSUE AND HOPE TO HAVE
THE PROBLEM RESOLVED IN A TIMELY FASHION. THANK YOU FOR YOUR
PATIENCE.
Enough fun for tonight ;-)
Cheers,
Tom
--
+----------------------------------------------------------------------------+
* Tom Yoksas UCAR Unidata Program *
* (303) 497-8642 (last resort) P.O. Box 3000 *
* yoksas@xxxxxxxxxxxxxxxx Boulder, CO 80307 *
* Unidata WWW Service http://www.unidata.ucar.edu/*
+----------------------------------------------------------------------------+
