#openstack-cyborg: openstack-cyborg
Meeting started by zhipeng at 15:00:57 UTC
(full logs).
Meeting summary
- Roll Call (zhipeng, 15:01:08)
- Howard (zhipeng,
15:01:15)
- Rushil (crushil,
15:09:07)
- Sydney Summit slide review (zhipeng, 15:11:27)
- Placement-Cyborg Interaction Discussion (zhipeng, 15:19:02)
- https://github.com/jaypipes/articles/blob/master/openstack/placement-queens-update.md
(zhipeng,
15:20:18)
- Jay Pipe's writeup on Placement (zhipeng,
15:20:46)
- http://specs.openstack.org/openstack/nova-specs/specs/queens/approved/ironic-driver-traits.html
(zhipeng,
15:21:23)
- https://review.openstack.org/#/c/504531/
(zhipeng,
15:21:40)
- first one is about Ironic report/accept trait
with Nova, second one is about trait config in Ironic (zhipeng,
15:22:17)
- http://specs.openstack.org/openstack/nova-specs/specs/queens/approved/virt-add-support-for-vgpu.html
(zhipeng,
15:23:00)
- https://github.com/kubernetes/community/pull/782
(zhipeng,
15:23:35)
- k8s discussion on resource class for
accelerators (zhipeng,
15:23:48)
- 1. We need to define CUSTOM_ traits for sure,
but also some additions to os-trait lib on FPGA if there are
any (zhipeng,
15:25:43)
- 2. other than the reporting from cyborg-agent
to the Placement API, we need (a) the driver also support trait api
and (b) cyborg-api support trait (zhipeng,
15:26:41)
- 3. There is a caveat for Cyborg on resource
provider modeling for accelerators which is we need to find a way to
model different types of accelerators in a unified way (zhipeng,
15:27:44)
- 3a. Traits and resource class are used to
define qualitive and quantative side of resource, however we might
need a third kind to descirbe accelerators, because only using
traits and ResourceClass means we will have a lot them for different
accelerators (zhipeng,
15:29:17)
- 3b. We might need to consider using something
like capability to describe the accelerator resource. For example
FPGA and GPU both could be used for ML, a user might just want a VM
to run a ML workload with the help of an attached accelerator
(zhipeng,
15:30:35)
- 3c. This means we need to enable user to be
able to select one capable resource instead of identifing very
specifically which resource they want (zhipeng,
15:31:29)
- ACTION: crushil to
work on a new spec on cyborg driver trait support, and zhipeng will
work on a new spec on api trait extention (zhipeng,
15:46:37)
- Generic Driver progress update (zhipeng, 15:54:14)
- ACTION: shaohe_feng_
and crushil collaborate on the driver development for backend
support (zhipeng,
15:58:09)
- ACTION: zhipeng to
send meeting invites for monthly deep dive (zhipeng,
16:03:39)
- patch discussion (zhipeng, 16:03:46)
- https://review.openstack.org/#/q/project:openstack/cyborg+status:open
(zhipeng,
16:04:05)
- AoB (zhipeng, 16:05:38)
Meeting ended at 16:09:54 UTC
(full logs).
Action items
- crushil to work on a new spec on cyborg driver trait support, and zhipeng will work on a new spec on api trait extention
- shaohe_feng_ and crushil collaborate on the driver development for backend support
- zhipeng to send meeting invites for monthly deep dive
Action items, by person
- crushil
- crushil to work on a new spec on cyborg driver trait support, and zhipeng will work on a new spec on api trait extention
- shaohe_feng_ and crushil collaborate on the driver development for backend support
- shaohe_feng_
- shaohe_feng_ and crushil collaborate on the driver development for backend support
- zhipeng
- crushil to work on a new spec on cyborg driver trait support, and zhipeng will work on a new spec on api trait extention
- zhipeng to send meeting invites for monthly deep dive
People present (lines said)
- zhipeng (103)
- crushil (39)
- shaohe_feng_ (17)
- openstack (3)
- zhuli (1)
Generated by MeetBot 0.1.4.