Using nagios for multiple data centers

[Matthew Litwin]

On 3/31/09 3:01 PM, "Marc Powell" <marc at ena.com> wrote:

> 
> On Mar 31, 2009, at 4:11 PM, Matthew Litwin wrote:
> 
>> I would like to pose a more advanced question to everyone here on
>> how I might use our current implementation of nagios to monitor
>> multiple data centers without having to set up multiple nagios
>> servers for each. The reason for doing is to have the ability to
>> view all nagios alerts from a single incidents page, generate
>> integrated graphs and reports, and reduce the number of servers in
>> general. However, while doing this, I want to also be able to be
>> able to view the just the alerts that are relevant to a single data
>> center at a time, thus limiting the view to just those alerts.
>> 
>> The function of our data centers will be that one is a mirror of the
>> other and will act as failover. While we need to make sure both are
>> up and healthy, one will be in an active state and one will be in an
>> standby state, and our NOC will need to be able to focus on just one
>> or the other. Thus, aside from assigning each datacenter with active
>> and standby state, one concern is to be able to view just the alerts
>> for the hosts and services relevant to a single data center. This
>> means having a main screen like nagios has for alerts, but one for
>> data center A and the other for data center B. I was thinking that I
>> could delineate which hardware was part of which data center by
>> using hostgroups, dependencies, or a mix of both. The point of this
>> would be to be able to view and report data from hosts and services
>> from a single data center as well as all of them.
>> 
>> Are there any documented methods for doing this that you know of?
> 
> While hostgroups are useful to group related devices together, the
> view restriction you want is a natural extension of configuring
> authorization for the CGI's (documented). We use it to limit views by
> State where we have hundreds of devices and dedicated Helpdesks. You
> can create specific user contacts or role contacts that can be
> associated with the hosts they should see. I would suggest the role
> account path, one per data center with a master contact that sees
> everything. Associate contact 'DataCenterA' with all hosts at Data
> Center A. Associcate contact 'DataCenterB' with all hosts at Data
> Center B. Associate the contact 'Master' with all hosts (or use
> authorized_for_* options in cgi.cfg). When someone logs in with one of
> those contacts usernames, they'll only be able to see hosts/services
> for which they are a contact. You can continue to use more personal
> contact addresses for notifications if you like. If you don't want
> them logging in with that personal contact name, don't create an
> htaccess account for it.
> 
> --
> Marc
> 

Marc, thanks for the suggestion. While the role contacts is certainly useful
as you described, I don't necessarily want to permanently limit users to not
seeing hosts at all.

Another way to think about what I am looking for is that I want to have a
way to designate servers to be either "active" or "standby" servers. All
servers have full monitoring set up but only "active" servers will appear in
the "active" servers' alerts view that the NOC sees. I want to create some
organizational objects that I can easily send as arguments to the CGIs to
create some some set alert views.

Does that make sense? Is this something that is a documented function of the
CGIs as well?


------------------------------------------------------------------------------
_______________________________________________
Nagios-users mailing list
Nagios-users at lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/nagios-users
::: Please include Nagios version, plugin version (-v) and OS when reporting any issue. 
::: Messages without supporting info will risk being sent to /dev/null