Notes from Distro OpenStack meeting 2/19/2019...

We went through all of the items in the list and downgraded many to “At Risk” for the coming release.  These are primarily in Nova, and will not make it into this release unless some new action is taken on behalf of the StarlingX community to influence them in upstream.

I will raise this overall risk to the TSC agenda tomorrow.

Details below and in colour at [0].

Bill…

[0] Distro OpenStack Tracker: https://docs.google.com/spreadsheets/d/1udAtEpQljV2JZVs-525UhWyx-5ePOaSSkKD1CS27ohU/edit?usp=sharing

Summary

       - At Risk: 11 items

       - No Plan: 13 items
       - Likely: 12 items
       - Certain: 4 items

Nova Features (9 at risk, 1 no plan)

Live Migration Patch w/ Pinning (numa_topology): At Risk

       - Nova (Artem?)

       - need to check where we’re at with this – at least we’re seeing some progress on this

       - monitor to see if it’s going to land soon

Shared/Dedicated vcpus on the same host or standardize CPU resource tracking: At Risk

       - Nova (Jay Pipes)

       - this hasn’t been active for ~9 weeks – so it’s At Risk for sure

Shared VCPU Extension: At Risk

       - Nova (TBD) – same as previous one

vCPU model: No Plan

       - Intel, unassigned

       - no chance of this being done for this release now

Clean Up Orphan Instances: At Risk

       - Yongli

       - he says this is lower priority – has time (but isn’t working on it right now)

       - lowering status to At Risk – likely to suffer same lack of attention in community

Server Group Show Enhancement: At Risk

       - Yongli

       - code has been posted, is actively working on it

       - not this week, but maybe next week (no Core reviewer yet)

       - Frank asked Yongli how he thinks we can get Core attention on this

       - can we get Alex Xu to +2 this? – Yongli will talk to him

Numa Topology API: At Risk

       - Yongli

       - this is in the same boat as the previous one

       - Yongli will talk to Alex Xu about this too 

Add support for emulated virtual TPM: At Risk

       - Chris Friesen, Certain

       - this will be taken over by Chris F, will try to get it in

       - downgraded this to At Risk

Disk Performance Fixes: At Risk

       - Jack Ding

       - suffering from lack of reviews by Cores

       - Frank will meet with Chris to see if he can help

       - get him to +1 it & say who would be the right Core to chase

Extra Spec Validation: At Risk

       - Jack Ding

       - same comment as previous feature

Nova Refactor (1 Likely)

PCI Interrupt Affinity: Likely

       - Yong Hu

       - planning to have code posted for review by end of next week (March 1st)

Nova Config (1 Likely)

RPC Timeout Handling: Likely

       - Frank

       - planning to assign this after the container cutover (target is March 31^st)

Nova Features – Test Only (need plan from test team)

Replace SR-IOV/PT best effort scheduling policy with Queens feature

Replace vswitch affinity with Rocky feature

DB purge

       - need Ada to provide plan for completing these test activities

Nova Bug Fixes (10 No Plan, unstaffed)

rbd disks: convert from source format to raw Intel to fix upstram

Migrate code to use request_spec for server groups

Specify migration type in db filter for anti-affinity migration

Do not bypass scheduler when using forced hosts

Exclude current instance when checking server group live migrations

Live block migration fails following auth token expiration

Drop support for forced hosts for live migrate

Unsquash: primary: server-groups best-effort and group size

Unsquash: primary: live migration driver fixes

Unsquash: Fix stale RequestSpec instance numa topology for live-migration

       - currently these 10 bug fixes are unstaffed, will not make it into the release

Neutron Features (5 Likely – but need updates)

Network Segment Range Management (top Neutron priority)

       - Kailun, likely – need update

Populate binding levels when concurrent ops fail (da31d0e)

       - Kailun, likely – need update

DHCP agent rescheduling / rebalancing

       - Kevin Smith, likely – need update

L3 agent rescheduling / rebalancing

       - Kevin Smith, likely – need udpate

Modeling the provider networks in sysinv

       - John Kung, likely – need udpate

Neutron Bug Fixes (2 At Risk, 2 Likely, 1 Certain) – need updates

Prevent DHCP from processing stale RPC messages (fd6cfc)

       - Kailun, Certain – need update

0039-dhcp-handle-concurrent-port-creation-error.patch (a38f89)

       - Kailun, At Risk – need update

0047-wsgi-prevent-accepting-socket-without-a-gr.patch (8e72491)

       - Le Huifeng, At Risk – need update

0062-dvr-force-admin-state-update-before-distri.patch (edf9731)

       - Matt Welch, Likely – need update

0065-dvr-do-not-create-agent-gateway-ports-unle.patch (2857911)

       - Enyinna, Likely – need update

Horizon Features (2 Likely, 2 No Plan)

CSRF AngularJS Fixes: Likely

       - Yan Chen

       - Yong to check, he thinks it may have been merged already

Custom Confirmation Messages: Not Required

       - Yan Chen

       - under discussion now, may decide we don’t need this

       - post-meeting, got update that this has been confirmed

Configurable Auto Page Refresh: No Plan

       - Intel

       - currently no plan, unstaffed

Auth Plugin Extensions: No Plan

       - WR

       - currently no plan, unstaffed – Bill to check with Dariush on this

Move Items to stx-gui: Likely

       - Kristine Bujold

      - in progress, some work to be done after Stein cutover (re: port forwarding)

       - it’ll be done by mid-March, it’s not upstreaming, so no risk

Glance/Cinder Refactor (1 Likely)

Refactor StarlingX to make use of Cinder caching to replace glance-cache raw implementation.

       - Liang Fang, Likely

       - need to get an update from Liang on forecast – may have to reforecast based on not using Puppet; Liang may need help with that