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

20040914: gempak processing getting wedged...



>From: Gerry Creager n5jxs <address@hidden>
>Organization: AATLT, Texas A&M University
>Keywords: 200409141117.i8EBHanJ003745 LDM NEXRAD2

Hi Gerry,

>I've been doing a fair bit of Level II image generation on gemdata2 
>recently, and I'm seeing a lot of interprocess communications hangups 
>now.  I've a set of questions.
>
>1.  We're using the transfer of the site volume scan from hidden to 
>"normal" (\E$) to fire the script that does image generation for that 
>site.  Is there a better trigger?

I don't think so.  the final quadrant of a full volume scan contains
the 'E' indicator for end.  This, in turn, kicks off the renaming
of the hidden file so it is viewable.  At that point, the data should
be read to be processed in any way you want.

>1a.  When data's missing, the script failure rate skyrockets.  Great. 
>Makes sense, but should that script even be firing?

Since it is -currently- possible that a piece of the volume scan is
missing (more on this below), kicking off a script that requires
every piece of the volume scan to exist to run correctly can be
problematic.  A more correct/complet/heavyweight approach would be
for the script to somehow determine if the volume scan has any
missing pieces and then do the GEMPAK processing if it doesn't.
However, I have seen Chiz put up plots from volume scans that were
missing pieces, so maybe there is more to what is going on that
the simple view.

>2.  We're now seeing high load averages and resource exhaustion. Might 
>this be playing into the mix?

It could, yes.

>The system's a single cpu celeron at 2 
>GHz with 512 MB RAM.  I'm thinking a change to 1 GB (because I've got a 
>system with that in place available for this) might be the key.  Or at 
>least a help.

What kind of load averages are you seeing?  I would have thought that
a 2 Ghz machine with 512 MB of RAM should be sufficient.

>It's still early in Texas... and in DC by my body's clock... so I've not 
>heard a status on bigbird yet.

I am unable to ping, ldmping, or ssh bigbird at this time, so it looks
like it is still down.

Cheers,

Tom
--
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.