[dev.icinga.com #9117] Make features HA-aware

[icinga-migration]

This issue has been migrated from Redmine: https://dev.icinga.com/issues/9117

Created by dgoetz on 2015-04-20 12:13:28 +00:00

Assignee: (none)
Status: New
Target Version: Backlog
Last Update: 2016-08-16 10:25:34 +00:00 (in Redmine)

Backport?: Not yet backported
Include in Changelog: 1

---

Some features are already ha aware, I think Perfdata-Writer and perhaps others like Graphite-Writer needs to be aware also.

My environment for example:

• Two nodes icinga2 in master zone
• One node with the webinterface and pnp4nagios

Workarounds now:

• Both master feature perfdata is active, I need a script which gets the data from one host and purges the others and switches if the first one is down and handle recovery properly, so I do net mess up data.
• Only one node manual intervention is neccessary on both sides and I have gaps.

Best solution would be every host writes its own perfdata for only the executed and not the transferred checks and I collect all perfdata files. (Not sure if I configure now a suffix with the endpoints name to not get a file collision.)

---

Relations:

• relates #9117

[icinga-migration]

Updated by mfrosch on 2015-04-20 12:14:28 +00:00

Is this possible?

[icinga-migration]

Updated by mfrosch on 2015-04-27 08:13:14 +00:00

• Target Version set to 2.4.0
• Estimated Hours set to 2

We think about enabling enable_ha for all useful objects, PerfdataWriter, GraphiteWriter, GelfWriter, OpenTSBWriter

This will enable a similar handling like the IDO, only one of the Nodes in a Zone will write Perfdata. And in failover or re-configuration case, another node will do it.

In terms of pnp4nagios the user would be required to take care about the file shipping.

Please plan it for 2.4, we except just a simple implementation

[icinga-migration]

Updated by mfriedrich on 2015-04-27 08:25:55 +00:00

• Estimated Hours deleted 2

[icinga-migration]

Updated by mfriedrich on 2015-04-29 06:29:27 +00:00

• Target Version deleted 2.4.0

We will discuss this further next week. We'll take care about time estimates/target versions then.

[icinga-migration]

Updated by mwaldmueller on 2015-05-07 08:52:40 +00:00

I think enable_ha is necessary, especially for graphite and perfdata.

[icinga-migration]

Updated by essener61 on 2015-05-07 11:55:55 +00:00

I have the same Problem... Our enviroment is like dirk has decribed.

[icinga-migration]

Updated by mfriedrich on 2015-06-23 13:30:49 +00:00

• Target Version set to Backlog

[icinga-migration]

Updated by mfriedrich on 2016-03-09 10:13:57 +00:00

• Relates set to 11335

[icinga-migration]

Updated by friesoft on 2016-03-18 10:10:35 +00:00

Any news on this?

We are currently setting up an Icinga HA cluster using 3 Nodes. The IDO mysql combined with MariaDB Galera Cluster works great.

We now need something similar for a graphite cluster... or is the only option we've got that every instance logs into its own standalone graphite?

[icinga-migration]

Updated by mfriedrich on 2016-04-07 10:39:57 +00:00

• Subject changed from Make Perfdata-Writer high availability aware to Make features HA-aware

[icinga-migration]

Updated by striep on 2016-05-29 15:25:26 +00:00

Hi,

we are also eagerly waiting for this feature as we plan to set up a HA cluster for icinga2 also with galera cluster

[icinga-migration]

Updated by ziaunys on 2016-06-14 23:50:24 +00:00

+1

[icinga-migration]

Updated by DG3AZ on 2016-08-16 10:25:34 +00:00

Any news? :-)

André

[friesoft]

Any news? :) We would love to use this in combination with the new logstashwriter with an icinga2 cluster
With 2 nodes and a central elastic stack.

[dnsmichi]

The LogstashWriter is still undergoing review and has not been merged to release branches. In terms of this feature - please dive into the code and contribute a patch.