[Wien] ERRCLR file is not close and parallel-wien stops

[L. D. Marks]

Probably not the reason. There is a delay between when the child node does 
an action (or says it will) and when this shows up on the master node; 
this is set by nfs/automount/whatever you are running. Your "touch" is
probably forcing the OS to sync the file information by clearing up 
nfs action items that are currently buffered in a "to do" list the.

N.B., I like this; it might well be something worth implimenting more 
generally to overcome sleepy (even asleep) NFS issues.

N.N.B., check that you are running enough nfs daemons, e.g. ps -uroot | 
grep -e nfs. If you only have a few (4-8) increase this. Somewhere there 
is info about this, e.g. man -k nfs & nfsstat.

On Tue, 18 Apr 2006, Alessandro Mirone wrote:

> Hi,
> I was having this problem with lapw1para  on a cluster :
>  the child processes exit but the error file ( seen from the master
> node ) continues to be, during several second, a non empty
> error file. If the check happens before the updating of the file with a null 
> content
> the program stops with error althoutght no error is there.
>
> I empirically solved the problem adding a touch *error to the parascript.
> That works greatly, apparently such touch forces an updating
> of the buffers.
> But indeed I think that the real cause is the missing of a close statement
> in errclr
> Regards
>    Alessandro MIRONE
>
>
> _______________________________________________
> Wien mailing list
> Wien at zeus.theochem.tuwien.ac.at
> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
>

Note: if you have an old email address for me, please note that "nwu" has
been changed to "northwestern".
-----------------------------------------------
Laurence Marks
Department of Materials Science and Engineering
MSE Rm 2036 Cook Hall
2225 N Campus Drive
Northwestern University
Evanston, IL 60208, USA
Tel: (847) 491-3996 Fax: (847) 491-7820
email: L-marks at northwestern dot edu
http://www.numis.northwestern.edu
-----------------------------------------------