Problem with check_oracle

[Hans Engelen]

On 9/30/05, Andreas Ericsson <ae at op5.se> wrote:
>
> > The biggest problem I have had with check_oracle however is when it is
> run
> > by the Nagios process. I have found this bit to be extremely cumbersome
> on
> > my setup. I kept getting errormessages about the client not being able
> to
> > set up the NLS environment which ... quite frankly proved to be a
> serious
> > problem to tackle. Best I can figure it has something to do with the
> > NLS_LANG (or was it LANG) environment variable and others not being
> properly
> > setup (in fact I suspect no environment variables get set) when run via
> > Nagios.
>
>
> It's the LANG environment variable. NLS_LANG has no special meaning.
>
> Nagios uses the library wrapper popen() to execute other programs (for
> now, anyways). popen()'ed children inherit the environment from the
> parent (but doesn't read any of the profile or rc-files; sh -c is called
> in to execute the program in question). At least on Linux and *BSD,
> although I can't imagine those systems going through the extra trouble
> of copying the environment data unless it was in POSIX.

 That sorta fits in then with what was going on. Depending on the client
version you have installed (and I took the version 10 client of course,
grumble) those variables have to be set though or your check_nagios goes
horribly wrong. I think I even went sofar as to try setting them from
withing check_nagios itself at one point before getting to the sqlclient
part but ...
 Is there any ... ehm ... approved solution for that ?
 Cheers,
Hans
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://www.monitoring-lists.org/archive/users/attachments/20050930/45314132/attachment.html>