On Thu, 02 Dec 2010 11:44:07 +0100, Carl Melgaard wrote:
Im monitoring a cluster-setup, a/b-nodes + cluster ressource:
10.83.13.12 bskidmsrv001 # COMMENT:"SLES11cluster-res" \
prefer
10.83.13.10 bskidmsrv001a # COMMENT:"SLES11a" \
ldap ldaps ntp ntpd \
badntp:1:3:3 prefer \
NOCOLUMNS:ldap,ldaps
10.83.13.11 bskidmsrv001b # COMMENT:"SLES11b" \
ldap ldaps ntp ntpd \
badntp:1:3:3 prefer \
NOCOLUMNS:ldap,ldaps
ldap + ldaps will always be red on one of the nodes, so I've made a
combotest to insert on the cluster-ressource:
bskidmsrv001.ldap=(bskidmsrv001a.ldap + bskidmsrv001b.ldap) == 1
bskidmsrv001.ldaps=(bskidmsrv001a.ldaps+bskidmsrv001b.ldaps) == 1
This works on - the green status shows up on the cluster-resseource host
- BUT I still get alerting on all the hidden red-tests, and I can still
acknowledge all of the tests, I thought I hid away with NOCOLUMNS.
How do I achieve this - ie. suppressing the red alerts on the nodes
where ldap/ldaps is not running?
Xymon has a problem with clusters. It should be possible to tell Xymon
that "these nodes are in a cluster, so don't alert me unless at
least N of them are down at the same time". Right now, it isn't.
So if you want to eliminate the red status on the individual nodes
in the cluster, you cannot do that.
If it's only the alerts you want to stop, then you can configure
no alerts to be sent for the individual nodes - only send alerts
for the cluster status (bskidmsrv001.ldap or .ldaps).
Regards,
Henrik