Yesterday puppetdb (nitrogen) was unavailable due to ganeti outage (https://wikitech.wikimedia.org/wiki/Incident_documentation/20170629-ganeti) and that failure was impacting puppet runs across the fleet. It looks like recent versions of the puppetdb terminus support multiple server_uris and are capable of failing over among them: https://docs.puppet.com/puppetdb/latest/puppetdb_connection.html#main-settings
Description
Description
Status | Subtype | Assigned | Task | ||
---|---|---|---|---|---|
Open | None | T294906 Puppet Improvements | |||
Open | None | T169318 Use multiple puppetdbs on puppet masters |
Event Timeline
Comment Actions
Im curious how puppetdb failed? do you rember? As the postgress write master is always on the primary puppetdb server im not sure we would get much of a win here. unless we also spun up a seperate postgress database