Lwood-20151108

November 9, 2015 at 22:27 · Filed under Lwood, OpenStack

Introduction

Welcome to Last week on OpenStack Dev (“Lwood”) for the week ending 8th November 2015. For more background on Lwood, please refer here.

The Mitaka Summit induced quiet spell last week was more than compensated for this week :)

Basic Stats for week 2nd to 8th November 2015 :

  • ~686 Messages (up about 155% from last week!)
  • ~204 Threads (up around 80% from last week)

Notable Discussions

Mitaka Release Schedule Published

Doug Hellman posted the Mitaka release schedule noting among other things there are only five weeks between Feature Freeze and Final Release instead of the usual six.

Exposing hypervisor/hardware details to users

Tony Breeds kicked off an interesting discussion about how best to expose hypervisor information to end users.  While seen by a few as a controversial idea there was far more assent than dissent and a way forward for something useable in ‘N’ and quite possibly Mitaka.

The most favoured approach seems to be to provide a capabilities API that abstracts the underlying information in hypervisor (or hardware) independent manner – a bit more heavy lifting to design and implement in the first instance, but a lot more future proof and allows a useful degree of hardware/hypervisor independence (the world isn’t all x86 after all…)

Changes to release process

Doug Hellman penned a few posts about changes to the release process for Mitaka.

The first dealt with communications – in summary the IRC channel for release related discussions has been renamed #openstack-release from it’s old name of #openstack-relmgr-office  Some behind the scenes tinkering should mean that anyone connecting to the old channel will be redirected.  The concept of “office hours” for the channel has now been removed – a recognition that most people have some form of scrollback or bot in use so that they can see messages posted while they weren’t logged in to the channel directly.

The second is more substantial – a move away from treating milestones as strict synchronisation points between projects, instead they are now intended to serve as reminders to the projects to have their own checkpoints.  Thus each project will be responsible for handling milestone tasks during the relevant weeks of the schedule, but at a big tent level there will no longer be an expectation to have all the tags and launchpad updates applied on the same day.

In a third email Doug outlines how similar changes are being made for stable releases and some in a fourth email, and finally in a fifth email some tools that might assist in keeping up.

Distributed lock managers at the Summit

Robert Collins posted a summary of conversations at the summit around adopting a common Distributed Lock Manager (DLM) While DLM is fairly deep developer level voodoo in most cases there are implications for operators as well as OpenStack moves in time to more unified solutions for this.

The plan put forward for now is to use an oslo abstraction layer named tooz – as Robert points out this doesn’t preclude using something else in special cases, but merely that they should be considered the exception not the norm.

It’s a longish thread that follows, but worth a read if DLM is on your radar and at the time of writing, still ongoing…

Oh and while you’re at it, take a moment to read this thread amusingly titled “[all][dlm] Zookeeper and openjdk, mythbusted” for a bit of comfort around the misconception that all of the above might mean you need to run a proprietary JVM…

Keep Juno alive for longer!

Tony Breeds kicked off a second thread worthy of a read – this one presenting a case for keeping the Juno release around for longer on the basis of its wide adoption and, as it would appear, a non-trivial number of customers that who are, for whatever reason, unable to contemplate a move to Kilo anytime soon.  

Post Mitaka Summit Summaries and Priorities

If you only have time to read one summary from the Summit, please consider making it this one by Doug Hellman as he neatly sets the stage for many themes that will be at the fore this cycle.

Beyond that, many projects and individuals posted summaries, recaps or priority lists following the Mitaka Summit and I have sought to link them all below, roughly in chronological order of their appearance on the list;

Midcycle dates and locations

A couple of midcycles (or lack thereof) were announced;

People and Projects

  • [rpm-packing] Alan Pevec and Jakub Ruzick were nominated for RPM packaging core by Haikel Guemar
  • [keystone] Steve Martinelli announced some changes to their cross project liaison personnel
  • [TripleO] Greg Haynes proposed Ian Weinand as core review on diskimage-builder
  • [ceilometer] Gord Chung proposed Rohit Jaiswal to be added to Ceilometer core
  • [networking-onos] Vikram Choudhary proposed Albert Dongfeng and Ramahjaneya Reddy Palleti as new cores for the networking-onos project
  • John Garbutt proposed adding Sylvain Bauza and Alex Xu to nova-core
  • [NFV][telco] Marc Koderer advised he would be stepping down from his TelcoWG core team membership
  • [neutron] After mooting such a move at the summit, Edgar Magana confirmed his decision to step down from Neutron core

Comments are closed.