Hi Kris,
in analysis.cfg you can only set thresholds for disks which are reported by the client and thus are already visible in the disk and inode column of your Xymon server.
If there is no USB drive and no graph, then you don't need to mess up with server side config.
By default the Xymon client will only provide info about local disks (which makes sense, as the remote mounted disks should be monitored at the source and not at dozens of connected servers), and also only for fixed mounted disks to avoid alerts if a removable media gets disconnected.
If you want to change this you have to modify the linux client script (but as pointed out, ensure the output as Posix standard to not confuse the server side evaluation!).
So it is no problem to remove the "l" from "df -Pl", but not the "P". You can also remove the "-x $EXCLUDES" w/o problems.
Does a "df -P | grep USB" show your USB drive?
If yes, modify the client script and test by "./xymonclient-linux.sh | grep USB". This also should show your USB drive.
If not, the disk is not properly mounted.
Norbert
Am Di., 29. Nov. 2022 um 17:11?Uhr schrieb Kris Springer <user-c2caa0a7a8d5@xymon.invalid>:
Continued thanks for the responses.? I did try commenting
everything out
of the analysis.cfg except for the default DISK * line, but still no
joy.? I don't think analysis.cfg is causing the USB to be
ignored.? The
xymonclient-linux script has been set back to it's original state.
Kris Springer
On 11/29/22 08:28, Adam Thorn wrote:On 29/11/2022 01:21, Kris Springer wrote:
Thanks for pointing me in the direction of the client script.? I
messed around with it a lot
I'd advise against editing the xymonclient script unless you're
very
careful, because...
So then in the script with all the df stuff commented out, I
had it run
df /dev/sda
...putting this in the client script will break server-side
processing
because...
The results of the script show the USB, but nothing shows up on
the
Server Disk page except this error.
Expected strings (Capacity and Mounted) not found in df output
..one of the df arguments specified in the original client
script is
"-P" which specifies "use the POSIX output format". Compare the
header
row generated by df with and without -P:
$ df | head -n1
Filesystem???? 1K-blocks??? Used Available Use% Mounted on
$ df -P | head -n1
Filesystem???? 1024-blocks??? Used Available Capacity Mounted on
xymond is expecting the latter format, so by manually hacking the
client script and changing the arguments, the message sent to the
server is no longer in the format it expects - which is what's
leading
to that "Expected strings ... not found" message, I think.
I'd suggest putting the original client script back in place and
debugging from there. Perhaps a good place to start would be with a
very simple analysis.cfg with just two lines:
DEFAULT
? DISK * 90 95
and nothing else. If your disk then shows up in the 'disk'
report, the
problem is somewhere in your live analysis.cfg. I'm not immediately
sure when analysis.cfg is reread - I can't see any mention in the
documentation, but I may not be looking in the right place.
Adam