Nagios Support Forum

tmcdonald or abrist,
how many nagiosfusion-servers I need to pelace my current solution? Now I have 1 Master-server in Head-office and 4 Slave-servers in regions.

whitest wrote:tmcdonald or abrist,
how many nagiosfusion-servers I need to pelace my current solution? Now I have 1 Master-server in Head-office and 4 Slave-servers in regions.

How many total hosts and services do you have across those machines? Really since Fusion is only displaying data and not running checks it should not matter and 1 Fusion machine will suffice, but if you have a crazy amount (like 1 million objects) it might get sluggish.

Luckly there are not crazy amount =)
It wiil be 7-8 regions. 2-3k services on each.
So, I need I Fushion as Central desktop and 7-8 Nagios Cores in regions?

But with this setup you would loose the feature of the central configuration.
You have to maintain 7-8 nagios instances + nagiosfusion. Is this the road you want to hit?

klajosh2 wrote:But with this setup you would loose the feature of the central configuration.

Indeed. Mod_gearman currently is the only solid way to really run real distributed checks with 1 central config management server. Otherwise, you will need to maintain multiple separate configurations. Though, with enough forethought, a tool like puppet/chef and some clever scripting could make the management of multiple configs a breeze.

klajosh2 wrote:But with this setup you would loose the feature of the central configuration.
You have to maintain 7-8 nagios instances + nagiosfusion. Is this the road you want to hit?

My client wanted to have sepatare nagios for each region for active checks, notifies and dashboard (for local stuff). Central Nagios checks services only in head-office and recieves check-results from all region-nagioses. Central Nagios displays all network view for stuff in head-office.

I used scripting with Rsync and Passwordless SSH (in /etc/init.d/nagios) to acheive central configuration. Config files has been organized accordingly.

Going back to the original point of contention - I'd argue that mod_gearman is perfectly stable. We have many clients running gearman workers that are processing as many checks as your entire environment put together. It has proven pretty reliable for us. If you're experiencing mod_gearman crashes I suggest troubleshooting there and resolving that. As abrist said though - Fusion may be an even more appropriate solution for you depending on your use case.

jdalrymple wrote:Going back to the original point of contention - I'd argue that mod_gearman is perfectly stable. We have many clients running gearman workers that are processing as many checks as your entire environment put together. It has proven pretty reliable for us. If you're experiencing mod_gearman crashes I suggest troubleshooting there and resolving that. As abrist said though - Fusion may be an even more appropriate solution for you depending on your use case.

Does mod_gearman support Nagios 4.0 - Nagios 4.1 properly ?
I realy many times experienced orhanted checks in mod_gearman =(

Yes - but it's version dependent. Described here.

jdalrymple wrote:Going back to the original point of contention - I'd argue that mod_gearman is perfectly stable. We have many clients running gearman workers that are processing as many checks as your entire environment put together. It has proven pretty reliable for us. If you're experiencing mod_gearman crashes I suggest troubleshooting there and resolving that. As abrist said though - Fusion may be an even more appropriate solution for you depending on your use case.

Maybe this worth another thread: What is the best way to debug mod_gearman? I am asking because when crash comes I am usually not sitting in front of the server checking log files.
What I noticed - what is very strange actually - is that I do not see orphaned checks just checks are not executed. only thing helps is to restart main nagios process and gearmand service.
After this checks are starting. I am using nagios core 3.5.1 and gearmand 0.33.

klajosh