Discussion:
[Check_mk (english)] HA Monitoring cluster
(too old to reply)
Gerardo Ferreyra
2017-12-17 20:54:34 UTC
Permalink
Hi guys. I've read this:

https://mathias-kettner.com/cms_cma_cluster.html

I do not have a cmk rack.

So, my question:
Can this be done with CMK enterprise, installed on 2 physical nodes?
If yes, how?

Thanks!!
Robert Sander
2017-12-18 09:04:53 UTC
Permalink
Hi,
Post by Gerardo Ferreyra
https://mathias-kettner.com/cms_cma_cluster.html
I do not have a cmk rack.
Can this be done with CMK enterprise, installed on 2 physical nodes?
It can be done with the CEE cirt1 appliance.
Corosync/Pacemaker and DRBD are builtin and ready to use after a quick
configuration via web GUI.

Regards
--
Robert Sander
Heinlein Support GmbH
Schwedter Str. 8/9b, 10119 Berlin

http://www.heinlein-support.de

Tel: 030 / 405051-43
Fax: 030 / 405051-19

Zwangsangaben lt. §35a GmbHG:
HRB 93818 B / Amtsgericht Berlin-Charlottenburg,
GeschÀftsfÌhrer: Peer Heinlein -- Sitz: Berlin
c***@kohly.de
2018-01-31 20:31:54 UTC
Permalink
Hi,

I found that all local checks on a cluster node has a different output
than on a unclusterd node.

In example i monitor the signal strength of two vdr systems, one on a
clustered pve workstation and the other on an unclustered stand alone
system.
The cluster only includes the pve services like vms and network interfaces.

The output is as follows:

unclustered node:
OK - VDR plugin femon sgnl is at 60 %
clusterd node
OK - On node hv02.chaos.inc: OK - VDR plugin femon sgnl is at 73 %

The local plugin is the same on both nodes.

And now _The Question of all Questions_:
Is this the wanted behavior?

Hint: The answer is not 42. ;)

Br.
Kohly
Andreas Döhler
2018-02-01 21:52:42 UTC
Permalink
Hi Kohly,

where is the problem? If you cluster a local check then on the cluster
resource you see the originator of this check result.
In you case the check result was produced on hv02.chaos.inc and then
clustered on your cluster resource.
Is there any error message or unwanted output with the clustered checks?

With clustered local checks you have the possibility to decide if the end
result should be the best state or the worst state of all single states.
In you case this is not relevant as only one node is producing a local
check result. But with more nodes producing the same local check
it will help to build a cluster state for this local check.

br
Andreas
Post by c***@kohly.de
Hi,
I found that all local checks on a cluster node has a different output
than on a unclusterd node.
In example i monitor the signal strength of two vdr systems, one on a
clustered pve workstation and the other on an unclustered stand alone
system.
The cluster only includes the pve services like vms and network interfaces.
OK - VDR plugin femon sgnl is at 60 %
clusterd node
OK - On node hv02.chaos.inc: OK - VDR plugin femon sgnl is at 73 %
The local plugin is the same on both nodes.
Is this the wanted behavior?
Hint: The answer is not 42. ;)
Br.
Kohly
_______________________________________________
checkmk-en mailing list
http://lists.mathias-kettner.de/mailman/listinfo/checkmk-en
kohly
2018-02-02 05:31:36 UTC
Permalink
Hi Andreas,

it isn't a real problem, but isn't it abnormal that all but the local
checks produces a 'normal' output even on a clustered host?
i only wonder about the different behavior.

br
Kohly
Post by Andreas Döhler
Hi Kohly,
where is the problem? If you cluster a local check then on the cluster
resource you see the originator of this check result.
In you case the check result was produced on hv02.chaos.inc and then
clustered on your cluster resource.
Is there any error message or unwanted output with the clustered checks?
With clustered local checks you have the possibility to decide if the
end result should be the best state or the worst state of all single
states.
In you case this is not relevant as only one node is producing a local
check result. But with more nodes producing the same local check
it will help to build a cluster state for this local check.
br
Andreas
Hi,
I found that all local checks on a cluster node has a different output
than on a unclusterd node.
In example i monitor the signal strength of two vdr systems, one on a
clustered pve workstation and the other on an unclustered stand alone
system.
The cluster only includes the pve services like vms and network interfaces.
OK - VDR plugin femon sgnl is at 60 %
clusterd node
OK - On node hv02.chaos.inc: OK - VDR plugin femon sgnl is at 73 %
The local plugin is the same on both nodes.
Is this the wanted behavior?
Hint: The answer is not 42. ;)
Br.
Kohly
_______________________________________________
checkmk-en mailing list
http://lists.mathias-kettner.de/mailman/listinfo/checkmk-en
Andreas Döhler
2018-02-02 22:24:30 UTC
Permalink
Clustered Windows services will give you the same output with the running
on xy-host.
Only difference is that the local check put it in front of the normal check
result.
In the end it makes more sense this way as check_mk cannot handle
active-active cluster settings with the normal mechanic. But it is possible
with the local check clusters.

Best regards
Andreas
Post by kohly
Hi Andreas,
it isn't a real problem, but isn't it abnormal that all but the local
checks produces a 'normal' output even on a clustered host?
i only wonder about the different behavior.
br
Kohly
Hi Kohly,
where is the problem? If you cluster a local check then on the cluster
resource you see the originator of this check result.
In you case the check result was produced on hv02.chaos.inc and then
clustered on your cluster resource.
Is there any error message or unwanted output with the clustered checks?
With clustered local checks you have the possibility to decide if the end
result should be the best state or the worst state of all single states.
In you case this is not relevant as only one node is producing a local
check result. But with more nodes producing the same local check
it will help to build a cluster state for this local check.
br
Andreas
Post by c***@kohly.de
Hi,
I found that all local checks on a cluster node has a different output
than on a unclusterd node.
In example i monitor the signal strength of two vdr systems, one on a
clustered pve workstation and the other on an unclustered stand alone
system.
The cluster only includes the pve services like vms and network interfaces.
OK - VDR plugin femon sgnl is at 60 %
clusterd node
OK - On node hv02.chaos.inc: OK - VDR plugin femon sgnl is at 73 %
The local plugin is the same on both nodes.
Is this the wanted behavior?
Hint: The answer is not 42. ;)
Br.
Kohly
_______________________________________________
checkmk-en mailing list
http://lists.mathias-kettner.de/mailman/listinfo/checkmk-en
_______________________________________________
checkmk-en mailing list
http://lists.mathias-kettner.de/mailman/listinfo/checkmk-en
Loading...