Thank you Alan, you were right. The version on the nodes wasn't the same as on the master.
After installing the new version everything seems to work fine.
Olivier Paradis Béland
-------- Message d'origine--------
De: Alan Riley [mailto:[email protected]]
Date: jeu. 2003-12-04 11:15
À: [email protected]
Cc:
Objet: st: Re: Stata on a Beowulf Cluster
Olivier Paradis-Beland ([email protected]) wrote
> I am an admin of a Beowulf Cluster of computers which runs stata.
> Until recently everything worked well but since a few weeks, I have
> some problems reading STATA format databases.
>
> What is strange is that when I am on the master computer (i.e. the
> node with the hard drive an the local OS), I can read my ".dta" files
> with stata correctly.
>
> However, when I log on a remote node (sharing its OS and physical
> resources via network with NFS protocol with the master) I cannot read
> the same file.
>
> Here is the error message:
>
> use scf86_roc.dta file scf86_roc.dta not Stata format r(610);
>
> The problem is recurrent on each of the nodes except for the master
> node. I suspect there may be a problem with a STATA config file
> somewhere.
It is hard to guess what might be going wrong. There is no Stata
config file that would influence this. I think the possibilities are
(in no particular order)
- The data is somehow becoming corrupted as it is transferred to
the remote nodes
- The remote nodes are somehow using an older version of Stata than
is on the master computer
- The remote nodes are somehow accessing a different copy of the
data than the master computer
To help diagnose this, I would run the following four commands
on the master computer and on the remote nodes and compare the
output to see if it offers any clues about the problem:
. update query
. ls -l scf86_roc.dta
. checksum scf86_roc.dta
. !sum scf86_roc.dta
The first command will tell you whether the copies of Stata on the
master computer and the remote nodes are at the same update level.
The second command will tell you whether the date and filesize of
the dataset are seen as being the same on the master computer and
remote nodes. The third command will calculate a checksum based on
the bytes in the file in question; this will tell you whether Stata
thinks the files are the same on the master computer and remote nodes.
The last command also runs a checksum, but it shells out to Unix
to run it. This will tell you whether Unix thinks the files are
the same on the master computer and remote nodes.
If you continue to have problems, please contact Stata technical
services at [email protected] will attempt to further
diagnose what might be going wrong.
--Alan
([email protected])
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
<<winmail.dat>>
