[omd-users] Advise needed on monitoring CTDB cluster servers

Jaroslaw Nowak 0xfate at gmail.com
Thu Sep 11 19:29:19 CEST 2014


while realizing monitoring of cluster nodes you need to differentiate
between monitoring the host itself and the service it provides.

while in a scenario where you have a fixed number of independent nodes you
would realize host monitoring to each independent node on it's independent
unique IP address, monitoring it's basic functionality (CPU, mem, hdd,
database-, smb, whatever-service,  etc.) .

while in a scenario where you have identical hosts being booted after a
host fails replacing the failed one, the replacing node should, in the most
cases, be identical to the one being replaced (virtual environment, ideally
a clone). Thus the host monitoring shouldn't be realized different from the
first one.

when this is done you would configure cluster services (in wato) monitoring
the shared IP address,  and your special checks to its provided services
(get/put a file on a share, check DB connection, your special check here).

Regards,
fate
 Am 11.09.2014 18:52 schrieb "Chang, Jason" <jasonc at infinityward.com>:

>  I'm trying to find a best solution of monitoring CTDB cluster servers.
> When a node from a cluster goes offline, the ip gets taken over by other
> nodes to sustain it's service level. I think this would cause some weird
> behaviors. Anyone successfully implemented CTDB monitoring through check_mk?
>
> _______________________________________________
> omd-users mailing list
> omd-users at lists.mathias-kettner.de
> http://lists.mathias-kettner.de/mailman/listinfo/omd-users
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mathias-kettner.de/pipermail/omd-users/attachments/20140911/e54da036/attachment.html>


More information about the omd-users mailing list