I am working on a Project for a logstash fail-over cluster (using pacemaker) and I am looking for an OCF (http://linux-ha.org/wiki/OCF_Resource_Agents) resource agent for logstash. My goal is to have Geo-reduntant Active/Standby-Cluster.
As far as I read, common approaches for logstash HA would be:
- Permanently running both logstash instances and use a virtual IP to transparently switch-over if one instance does not respond.
However I am not only pushing my messages to logstash, I am using the JDBC-Input which pulls from a Database, therefore a constantly running second logstash would duplicate the messages and it would double the load on the database
- Use a message queue to do the complicated fail-over stuff, maybe near the database, to avoid duplication and have 2 logstash instances read the same queue.
Introducing a queue would be a big change in the design and i would avoid it, if it is possible. And I would end up with one logstash instance at the other site constantly pulling data from the active site over WAN and pushing it back after processing
Have you come across an resource agent for logstash (google resulted in nothing for me)? Or do you know any other useful tool for managing logstash from the outside?