Xymon Mailing List Archive search

FILE in analysis.cfg

2 messages in this thread

list Paul Root · Thu, 11 Apr 2013 15:20:32 +0000 ·
So I'm creating a log file that will only get updated about every 3 months. But I want to watch that and make sure that it does get updated every three months.

That part is easy. But, the file is only for our Linux machines, not our Solaris machines. Is there a way to tell  analysis that I want to look for this file, but if it doesn't exist, that's ok? Or is there something like CLASS but for machine type?

Oh, and I'm also monitoring a handful of linux machines for others outside my group, that won't have this file. And if I just put in the FILE line, it's going to tag those machines with file doesn't exist, if I put it in HOST=*. And I don't really want to put it in, individually, for the 90+ machines I do have, and the ones I'm adding weekly.

The vast majority of my machines do have a 3 character start, so I guess I could use those...

Any better ideas?

While, I'm at it, is mtime> only in seconds or can I put a 'd' modifier for days on the end of it?
list Jeremy Laidman · Fri, 12 Apr 2013 12:27:57 +1000 ·
quoted from Paul Root
On 12 April 2013 01:20, Root, Paul T <user-76fdb6883669@xymon.invalid> wrote:
**

Is there a way to tell  analysis that I want to look for this file, but if
it doesn’t exist, that’s ok?
I don't think so.  But you can work around this in client-local.cfg, like
so:

file:`[ -f /path/to/file ] && echo /path/to/file || echo /dev/null`

This will simply send through a different filename in the client data, and
the server-side will act as if you haven't configured it at all, which I
think means it won't report anything.

Also, there seems to be an undocumented modifier "optional" for both FILE
and LOG clauses.  I haven't used this, but it seems you would just append
it to the end of the FILE line in analysis.cfg, like so:

FILE:/path/to/file red mtime<7776000 optional
quoted from Paul Root

While, I’m at it, is mtime> only in seconds or can I put a ‘d’ modifier
for days on the end of it?
Nope, the string after "mtime>" is fed straight into aton() before being
stored in the rule structure, so it can only be digits.

Cheers
Jeremy