Bug 1124670

Summary: [RFE][ceilometer]: Rebase partitioned alarm evaluation on taskflow
Product: Red Hat OpenStack Reporter: RHOS Integration <rhos-integ>
Component: RFEsAssignee: RHOS Maint <rhos-maint>
Severity: medium Docs Contact:
Priority: unspecified    
Version: unspecifiedCC: markmc, yeylon
Target Milestone: Upstream M3Keywords: FutureFeature
Target Release: ---   
Hardware: Unspecified   
OS: Unspecified   
URL: https://blueprints.launchpad.net/ceilometer/+spec/taskflow-alarm-partitioning
Whiteboard: upstream_milestone_juno-3 upstream_definition_approved upstream_status_needs-code-review
Fixed In Version: Doc Type: Enhancement
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2015-03-19 17:22:21 UTC Type: ---
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: ---

Description RHOS Integration 2014-07-30 04:04:44 UTC
Cloned from launchpad blueprint https://blueprints.launchpad.net/ceilometer/+spec/taskflow-alarm-partitioning.


In the long term, it could be a replacement for the PartitionedAlarmService. It uses the same basic idea: a master node keeps track of the alarms to be evaluated, but instead of distributing a slice of the alarms to each node, it just maintains a pool from which the other nodes can pick some work. This pool is managed by taskflow, for instance by using a multi-node Zookeeper configuration.

The election of the master node works as in the PartitionCoordinator.

In this service, the nodes are not "passive" (they do not wait for work to be given to them); they are "active" (they ask for work when they have nothing more to do).

Specification URL (additional information):