Hi J.C.,
I forgot to mention that I'm using the latest Terabithia RPM for my server
on RHEL 7, so I'm not able to deploy a source patch to my production server
very easily (at least from my understanding and ability)!
Would it be possible to put this into a test RHEL 7 RPM for me to test on
my production server? I'd really appreciate it!
Alternatively, I can work on getting a parallel Xymon server set up
(created from source) that data is propagated to (but no alerts are set up
for). This is something I've been meaning to do anyways (for other
reasons), and I can start working on this now to test this patch, if
necessary.
Let me know if I need to do the parallel server set up or if a test RHEL 7
RPM is possible.
Thanks as always, J.C.!!
--
Matt Vander Werf
On Sat, Oct 17, 2015 at 11:45 PM, J.C. Cleaver <user-87556346d4af@xymon.invalid>
wrote:
On Mon, October 12, 2015 6:20 am, Matt Vander Werf wrote:
Hello,
We've noticed that when total Swap being used is greater than the total
Swap on the machine in the memory column/page, Xymon reports back with 0%
for Swap usage and alongside it, it says, " - invalid data".
We've noticed this happening several times for us and I assume it's due
to
excessive swap usage that causes the memory/swap usage being reported by
the machine going completely out of wack...
Here's a recent example:
Memory Used Total Percentage[image: green]
Physical 10183M 11910M 85%[image: green] Actual
9945M 11910M 83%[image: clear] Swap
1759218M 3967M 0% - invalid data
This is fine and all and it's certainly not Xymon's fault in any way, as
it's just reporting what the machine is reporting (and the machine is
also
showing wacked out numbers for swap usage).
What I don't care for is that when this happens, the color/state for Swap
just goes Clear, which causes the memory page/column to just go Green for
the machine in question. This makes it very hard to know when this kind
of
thing happens and needs to be addressed (usually with a reboot of the
machine), as it just appears that everything is fine since memory is
showing Green. Often times, we don't even get any warning of excessive
swap
usage with any Yellow or Red alerts prior to this happening as it happens
so fast on the machine, too fast for Xymon to catch it.
Is it possible to make it so that this kind of situation causes the
color/state for Swap to go Red, as an indication that something is very
wrong and needs to be addressed, instead of going Clear? Or at least
makes
it go Yellow?
I'd greatly appreciate this being changed so it at least doesn't show
Clear! Any other suggestions for how to change this behavior are more
then
welcome!
Thanks for the help!!
Hi Matt,
Can you try the patch below? It should upgrade these to a Yellow status
result in this case (and in the parallel event that phys goes over 100%
too)... I think.
Regards,
-jc