AW: Verfy NFS link status

Hi Scott,

many thanks for your answers. I followed your advice and sent this message
"official".  The very best part in your answer was your question ".. why
mounting NFS within application programs ...". I can't find a reasonable
answer and we would surely avoid all these problems, when mounting and
unmounting would not be done in every started communication job. It was
probably the idea of the programmer to ensure the envrionment individually.
But this could be better done once after IPL as you stated below.
Nevertheless I'll try to implement a status check by using the STATVFS API,
which will then replace the perseverative unmount/mount operations. I would
appreciate If someone can give me an example of a RPG prototype for STATVFS
as I'm (not yet) familiar with C specifications.
I'll also have a look on the mentioned link to your utility with the QP0LROR
API. It could be helpful in some cases.
(I'm still struggeling with the iSeriesNetwork registration forms, but I'll
wangle it).

Thanks again for advice and help.
Kind regards,
GEFIS Gesellschaft für
Individual-Software mbH
Werner Noll
 

-----Ursprüngliche Nachricht-----
Von: Scott Klement [mailto:sk@xxxxxxxxxxxxxxxx]
Gesendet: Dienstag, 8. Juni 2004 17:25
An: Werner Noll
Betreff: Re: Verfy NFS link status



Hi again,

Sorry I forgot to reply to the second part of your message.

> If you are experienced with NFS, I have more questions on following
problem:

I have some experience with NFS, I'm not an expert, though.

> If I do an unmount (e.g. by RMVMFS command) after a sucessfull mount, the
> RMVMFS sometimes fails with CPFA09E (Object in use) and I couldn't find
out
> by whom it is used. The recommended usage of QP0LROR API is too much
> complicated for a "normal application programmer" like me.

Carsten Flensburg wrote a utility that uses this API to check for object
locks in the IFS.  I published this utility in my newsletter (Club Tech
iSeries Programming Tips.)   You might be able to adapt it to your
purposes.

You can read the article at the following link.  You'll need to register
with the iSeries Network (if you're not already) to read it, but a free
("Associate") account should work:
http://www.iseriesnetwork.com/resources/clubtech/index.cfm?fuseaction=ShowNe
wsletterIssue&ID=18306

> On the other hand, the RMVMFS command is not very "clever" as you can
> use it for any directory (not a mounted one!) and you will always get
> the confirmation message "CPCA1B1 File system or directory unmounted."
> What I have figured out is, when you execute the ADDMFS command e.g. 3
> times you have to execute the RMVMFS command also 3 times for removal
> (if it works).

Yes, that's standard with NFS.   If you mount it 3 times successfully,
you've mounted it over itself, so you've got 3 separate connections open
to the NFS server -- you have to unmount it 3 times to clear all 3 of
them.

> I assume that this problems can also depend on the different NFS target
> systems; in my case it's a Suse-LINUX.

In my case it's FreeBSD, but it appears to behave the same way.

Can I ask why you want to keep mounting and unmounting the file system?

Why not just keep it mounted all the time?

That's what I do...  I don't have to keep checking if it's mounted,
because it's always mounted.  I don't have to worry about unmounting it
when someone else is using the filesystem, because I never unmount it.  I
don't have problems with it being mounted more than once because the only
time I mount it is when I IPL the system, and the only time I unmount it
is when I take the system down.