Inductor
The Inductor consumes WorkOrders (rfc / configuration change) or ActionOrders
(start, stop, etc) from a queue by zone, executes them and posts a result
message back to the controller.
The account.Cloud.location is used by the controller to publish the order into a queue. The inductors consume,
does the work and publish the result back to the controller.response queue.
Here is the logical flow from CMS:
Inductor Details
The inductor is Java core with Ruby for control. The Java side is standard maven + spring,basically a Spring DefaultMessageListenerContainer using Apache Commons DefaultExecutor
to spawn either local chef-solo (for IaaS or non-managed via orders) or a remote
via SSH chef-solo execution.
The image below shows a logical view of the classes in com.oneops.inductor.
There is a ruby gem to simplify setup and control.
Source / Downloads
Control
An Inductor runs using Java jar with a several arguments. There is a gem or bash script to make easier.
Inductor gem: inductor help, start, stop, restart, status
or
Inductor control bash script located in the root dir of the repo: ./inductor start,stop,restart,status
Logs / Inductor Log Agent and Sink
The Inductor will put logs where the conf.dir’s log4j.xml specifies. The gem redirects to the relative log dir.
The inductor logs are shipped using logstash forwarder to backe end elastic search cluster.
The UI uses the daq api (Spring based) PerfController to get data.