Page 1 of 1

Re: [Nagios-devel] Load Balancing and Redundancy with Nagios

Posted: Wed Apr 03, 2002 11:00 am
by Guest
I just had another thought about the redundancy: I can also see a real
advantage for rotating the monitoring of each individual entry from one
location/host to another. This would allow you to do monitoring from
different "remote" locations, giving you a better view of what a customer
might see.

Kevin Benton

-----Original Message-----
From: Kevin Benton
Sent: Wednesday, April 03, 2002 11:51 AM
To: 'Subhendu Ghosh'
Cc: nagios-devel@lists.sourceforge.net
Subject: RE: [Nagios-devel] Load Balancing and Redundancy with Nagios


The problem with distributed monitoring is that it's a manual process to
distribute at this point. Automatic process distribution is a much better
model for scalability and redundancy.

For example, if each host has a comm parent that spawns monitoring
processes, the comm parents can talk to each other about what's happening.
If a comm parent dies, the remaining comm parents can pick up the slack.
That same comm parent could share configs between hosts so that they're all
synchronized. They could do that via a CRC check. When the CRC doesn't
match, the one with the newest config sends its files to the ones that don't
have it. For security reasons, you may choose to restrict that capability
to one or two monitoring hosts.

The monitoring web pages would be updated by the comm parents or a child
designed for that purpose, of course.

Kevin Benton

-----Original Message-----
From: Subhendu Ghosh [mailto:sghosh@sghosh.org]
Sent: Wednesday, April 03, 2002 11:40 AM
To: Kevin Benton
Cc: nagios-devel@lists.sourceforge.net
Subject: Re: [Nagios-devel] Load Balancing and Redundancy with Nagios


On Wed, 3 Apr 2002, Kevin Benton wrote:

> Is there a way to load balance Nagios on multiple hosts so that 1) the
load
> level is dropped on a per monitoring host basis, and 2) Nagios can have
> redundant monitors on multiple hosts? If not, is there a plan in the
works?
> In disaster recovery, it's critical to make sure that if host A goes down,
> that all it does is backed up by some other host (B). If monitoring hosts
A
> and B are both in the same subnet/facility on a multi-subnet/facility
> network, that can also create a problem. We work to make sure that we
place
> monitors in different locations as well as on different subnets.
>
> Kevin Benton
>

load balancing is achived today by a distributed monitoring.

Nothing prevents you from having redundant monitors on different hosts.
You could have one nagios installation monitor the presence of another and
upon detection of failure of the primary host, send itself a sighup with
additional config files added...

There are no currently defined methods for synchronizing configs across a
number of nagios hosts.

-sg





This post was automatically imported from historical nagios-devel mailing list archives
Original poster: ubhendu Ghosh [mailto:sghosh@sghosh.org