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

Re: 20050504: gempak cleanup script error



Gabe,

The _gf programs don't spawn a gplt or gf process because they are
liniked together in a single executable, and as a result, the "gf" you
have can absolutely not be from one of those scripts.

Since you have a gf process, you must have run a program that
isn't a _gf version, and so you must have gpend at the end
of those.

Steve Chiswell
Unidata User Support



On Mon, 2005-05-09 at 09:55, Gabe Langbauer wrote:
> Steve,
> 
> There are several scripts that don't use gpend...I talked with Chris
> Hennon, who wrote most of these scripts and he states "I use the program
> sfmap_gf (instead of sfmap) to
> create the gif files.  Programs that end in "_gf" do not spawn a gplt
> process, so you do not need to use 'gpend' to terminate it.  Those
> programs were created to make the scripts cleaner and run faster."
> 
> If that isn't true, I'll go ahead and put 'gpend' at the end of the
> scripts
> 
> Is it possible that the cleanup script is somehow interacting with a gf
> process that it shouldn't be?  It appears that the 'gf' process that is
> causing the trouble is always the one that is called following the cleanup
> script on the crontab.  
> 
> On 4 May 2005, Steve Chiswell wrote:
> 
> > Gabe,
> > 
> > The program "gf" is the GEMPAK X-window based gif driver. This is not
> > a cleanup script.
> > 
> > If you have "gf" processes as well as "gplt" processes hanging around,
> > then either:
> > 1) You are not running "gpend" to shut down the message queue and
> >    device driver.
> > 2) Your cron or other program that is producing gifs is crashing 
> >    before getting to the "gpend" command.
> > 
> > The "cleanup" script in the gempak distribution will remove old message
> > queues, and gplt processes associated with such problems, but if the
> > problems are frequent, then you need to look at what scripts etc you 
> > have that are being run to generate the problem.
> > 
> > Since the "gf" driver uses an X server, you must have an X server
> > available to connect to using the DISPLAY environmental variable.
> > Verify that you aren't having problems related to that. If you are
> > running crons, then typically you would be running a virtual X server
> > (xvfb) to use to generate displays, or use the "gif" DEVICE in your
> > script which does not use the X server.
> > 
> > Steve Chiswell
> > Unidata User SUpport
> > 
> > 
> > 
> > On Wed, 2005-05-04 at 09:38, Gabe Langbauer wrote:
> > > Steve,
> > > 
> > > You may remember me asking about our crashing system here at Ohio State a
> > > few weeks ago...I've coninued to monitor the situation and it appears that
> > > what is occuring is that the cleanup script "hangs" and takes up 99% of
> > > the cpu for hours on end, until the whole system just crashes. 
> > > 
> > > I'm only guessing that it's the cleanup script, the command name is "gf"
> > > which could be anything...but the time it is occuring seems to relate with
> > > the time the cleanup script should be running.  I even changed the cron
> > > time of the cleanup script to test, and the time of the crashes changed
> > > accordingly.
> > > 
> > > Is there any "fixes" to the cleanup script that I may be missing?  Is
> > > there any reason you can think of that the cleanup script may hang like
> > > this?  Any other thoughts?
> > > 
> > > My system has the following characteristics:
> > > Dell Xeon dual-processors with RedHat v. 3
> > > Gempak 5.7.3
> > > 
> > > Any help would be greatly appreciated!
> > > 
> > > Thanks,
> > > 
> > > Gabe Langbauer
> >