Revision as of 15:12, 31 March 2014

Weekly Ironic Project Team Meeting

If you're interested in bare metal deployments with OpenStack, please join us. This meeting is where we'll talk about the Ironic project, an evolution of the Nova Baremetal driver. Anyone is welcome to add items to the agenda below, or bring up their topic during the open discussion at the end of each meeting.

Meetings are held in the #openstack-meeting room on irc.freenode.net at 19:00 UTC on Mondays, and chaired by either Devananda or NobodyCam. The next meeting is scheduled for March 17th, 2014 (http://www.timeanddate.com/worldclock/fixedtime.html?iso=20140317T1900).

The related TripleO project has its team meeting on Tuesday at the same time, in the #openstack-meeting-alt room.

Agenda for next meeting

https://launchpad.net/ironic/+milestone/icehouse-rc1 Icehouse RC status]
Ongoing integration & testing work
Open Discussion:
- Functional/Integration testing of vendor drivers (Tempest?)
- Driver architecture for Chassis object (Allow vendors to define behavior for the Chassis object like Auto node discovery etc)
- What is the best way to organize common parts for Ironic and Ironic-python-agent. We have common functionality for disk partitioning and it is better to have a place where we can put common parts. Or we just need to have two independent implementations and then merge them if we want to store Ironic and Ironic-python-agent at the same repository?

Previous meetings

Logs from previous meetings can be found here.

@@ Line 16: / Line 16: @@
 ** Functional/Integration testing of vendor drivers (Tempest?)
 ** Driver architecture for Chassis object (Allow vendors to define behavior for the Chassis object like Auto node discovery etc)
-** Ironic-python-agent architecture
+** What is the best way to organize common parts for Ironic and Ironic-python-agent. We have common functionality for disk partitioning and it is better to have a place where we can put common parts. Or we just need to have two independent implementations and then merge them if we want to store Ironic and Ironic-python-agent at the same repository?
-*** procedural vs declarative API (What do declarative API and procedural API exactly mean? Why can't we have both those approaches simultaneously? In fact, we can have granular tasks such as "disk config", "raid config", hardcoded (explicitly declared) tasks such as "prepare" and "deploy" as well as data driven task flows (flow is given in request).)
-*** external CMDB (What if someone wants to use Ironic-python-agent w/o ironic itself? It does not even require any support except being able to send heartbeat messages to external serivce taking URL from kernel cmdline. )
-*** heartbeat (What is the format of heartbeat message? Agent IP, MAC, ram, cpu, jtag, SNs, etc. Which node's field is supposed to store all those data? It is supposed that ir-api will match node by those data and update agent IP if necessary.)
-*** hardware inventory  (Is it supposed that inventory will be implemented as ir-cond periodic task or it will be request driven or some kind of integration with ceilometer or whatever? What if the number of nodes is huge?)
-*** Reference agent driver: https://blueprints.launchpad.net/ironic/+spec/agent-driver. What features should the agent driver include? currently: UUID discovery based on MAC addresses, heartbeat to ensure agent is still alive, sending commands to REST API.
 == Previous meetings ==
 [http://eavesdrop.openstack.org/meetings/ironic/ Logs from previous meetings can be found here.]

Difference between revisions of "Meetings/Ironic"

Revision as of 15:12, 31 March 2014

Weekly Ironic Project Team Meeting

Agenda for next meeting

Previous meetings