From the man page for combo.cfg:
A simple test - e.g. "Web1.http" - results in the value "1" ifthe http" test for server "Web1" is green, yellow or clear. It yields the value "0" if it is red, purple or blue.
Can someone explain the thinking behind that for me?
The reason I ask is because we are trying to reduce the number of columns we display. Any significant outage causes all sorts of columns to start lighting up. This causes the width of the xymon display to become very wide and quite unmanageable.
So we decided to experiment with combo tests and noticed a couple things, yellow alerts do not cause the combo test to go red. Disabling (blue-ing) a test causes the combo test to go to red. I blue or disable tests so I will not see the alert, typically when one or more hosts is going down for a maintenance.
Perhaps I am trying to use combo test in a way it was not designed for? Before I go looking in the code, to perhaps change what colors cause the combo test to fire, I would like to understand the thinking behind the combo test, assuming I cannot change the colors from a config file?
~David
Den 04-03-2011 15:48, Gore, David W skrev:
From the man page for combo.cfg:
A simple test - e.g. "Web1.http" - results in the value "1" ifthe http" test for server "Web1" is green, yellow or clear. It yields the value "0" if it is red, purple or blue.
Can someone explain the thinking behind that for me?
I can try ...
yellow alerts do not cause the combo test to go red.
The rationale for not causing a combo-test to alert on yellow, is because it would be an escalation of the status (one yellow status would trigger a combo to go red = more serious than yellow).
Disabling (blue-ing) a test causes the combo test to go to red.
That one I cannot explain. It doesn't seem right.
assuming I cannot change the colors from a config file?
It's currently hard-coded which colors trigger a red combo-status and which ones do not.
If you do want to try out some changes, then look at the "xymond/combostatus.c" file (in 4.3.0). Line 270 or so has
if (*color == -1) return -1;
else return ( (*color == COL_GREEN) ||
(*color == COL_YELLOW) ||
(*color == COL_CLEAR) );Adding a test for COL_BLUE there should make it behave a bit more sane.
Regards, Henrik
On Mar 4, 2011, at 10:22 AM, Henrik Størner wrote:
Disabling (blue-ing) a test causes the combo test to go to red.
That one I cannot explain. It doesn't seem right.
I stopped using them, mainly for this reason. So can we expect this behavior to be changed? Is it already changed in 4.3.0? Thanks.
-- Rob McBroom <http://www.skurfer.com/>
Den 09-03-2011 17:03, Rob McBroom skrev:
On Mar 4, 2011, at 10:22 AM, Henrik Størner wrote:
Disabling (blue-ing) a test causes the combo test to go to red.
That one I cannot explain. It doesn't seem right.
I stopped using them, mainly for this reason. So can we expect this behavior to be changed? Is it already changed in 4.3.0? Thanks.
Not in 4.3.0, I just put the update into the development tree, so it will appear in 4.3.1 (once I decide that "enough" bug- fixes have accumulated).
The attached patch will fix it - applies on top of 4.3.0. This makes the set of colors that are considered error-colors configurable, e.g. "--error-colors=red,yellow" will do that.
Regards, Henrik
participants (3)
-
david.gore＠verizonbusiness.com
-
henrik＠hswn.dk
-
mailinglist0＠skurfer.com