Zone Configs from Icinga Director not appearing on 2nd Master

Problem: I have an Icinga2 Setup with Director. Using a Single Master, it looks like this:
Zone 1 - 2 agents
Zone 2 - 2 agents
Zone 3 - 2 agents

This works pretty smoothly, but we have the need for a 2nd master.
I have introduced the 2nd master Master2, and ALL zones get deployed to /var/lib/icinga2/api/zones and the master successfully loads the configs. However, when the 2nd master takes over the IDO, only host/service objects that exist in the master zone appear in Icinga Web.

Again, on master2, all of the other zones exist in /var/lib/icinga2/api/zones – so something isn’t right.

I have done a sync from master1 to master2 as suggested in another forum post with no avail:
`scp -r /var/lib/icinga2/api/packages/_api root@master2:/var/lib/icinga2/api/packages/_api

I have also removed the /var/lib/icinga2/api/zones and /var/lib/icinga2/api/zones-stage directories to force a resync with no avail.

Spot checking the md5sums on some of the hosts.conf files doesn’t show anything amiss.

Give as much information as you can, e.g.

I know our versions are a little dated, but due to the nature of our environment, I cannot update without a major outage impacting customers. If you believe that an update to any of the packages will fix a known bug, please show me the update notes where it is addressed so I can kick it to change management. Otherwise, I just assume something is merely misconfigured on my end :slight_smile:

  • Director version (System - About): 1.8.1
  • Icinga Web 2 version and modules (System - About): 2.7.3
  • Icinga 2 version (icinga2 --version): icinga2 - The Icinga 2 network monitoring daemon (version: 2.13.0)
  • Operating System and version: CentOS 7 3.10.0-1127.8.2.el7.x86_64
  • Webserver, PHP versions: php7.1 / apache2.4.6

Any help would be greatly appreciated :slight_smile: – Again – I’m assuming this isn’t a bug a just a simple misconfiguration.

After a night of sleep, I revisited my sanity checks and still don’t see anything out of the ordinary. The team started using scheduled downtimes and had some issues, so there was an emergency upgrade to Icinga2 2.13.0 this morning.

Any other thoughts?

Edit to add – took the time to update incubator and director while I had an outage.

Bump – still looking for answers :slight_smile:

I did notice discrepancies in the configs between the HA Master Blog Post and the HA master documentation.

Blog post has the original master connecting to the 2nd master, while the documentation has the 2nd master connecting to the original master.

I have a maintenance window tonight to swap the config (currently I follow the blog post and have the original master connect to the 2nd). Tonight I will try having the 2nd connect to the original.

Of course, this might just be preference and not matter, but I’m desperate for anything at this point.

Not sure if it was the emergency upgrades or the swapping of the config files, but it looks to be working now.

We had an issue where passive checks (and automated Acks via scripts) executed via the API stopped working if the calls were made against the node that is not the active master, so we are looking into this.

1 Like