Kubernetes Weekly Community Meeting

videos

#21

Sep 20, 2018

  • Moderators: Jonas Rosland [SIG-ContribEx]
  • Note Taker: Josh Berkus, Jaice Singer DuMars and Jorge Castro [SIG-ContribEx]
  • [ 0:01 ]** Release Updates**
    • Current Release Development Cycle [Tim Pepper ~ 1.12 Release lead]
      • We’re down to probably 1 issue pending fix and test soak
      • Key upcoming events barring any new test issues:
        • Friday Sept 21 “cherry-pick deadline” (non-event given master hasn’t yet thawed)
        • Friday Sept 21 cut RC2 and built rpms/debs
        • Monday Sept 24 last release-1.12 branch fast-forward from master branch
        • Monday Sept 24 thaw master branch
        • …final soak: cherry picks only for absolutely critical show stopper bugs…
        • Thursday Sept 27 release
    • Patch Release Updates
      • 1.9.10 released Aug 3
      • 1.10.8 released Sept 15
      • 1.11.3 released Sept 11
  • [ 0:03 ] SIG Updates
    • SIG Cloud Provider [Chris Hoge] (confirmed)
      • https://docs.google.com/presentation/d/186rAa3cNCBOA2GBmFdNvBI_Ko4inaQYHEX2CFA4TSqs/edit#slide=id.p
      • Looking at cloud providers in core code, like Google, AWS, etc.
        • Don’t want more in the upstream code
        • Want level playing field
        • In the process of moving providers to plugins
        • Minimum requirements to add a provider, is it documented? Do they post results to testgrid? So that users have some assurance that they’ll have a positive experience.
      • In 1.12:
      • In 1.13:
        • Continue work to move in-tree providers
        • Document external provider usage, need to make sure all requirements etc. are documented. Installation process is more complicated, for example.
        • Maybe moving provider SIGs into SIG-CP, likely to take longer.
        • Make sure all providers have conformance test results
      • Collaboration
        • SIG-Docs doesn’t want to be in charge of provider docs, that should be up to SIG-CP
        • Working with Cluster Lifecycle on install/upgrade
      • Q: should provider repos be in the Kubernetes org?
        • A: the decision on whether providers should be part of the kubernetes org is an Arch decision, it’s a question of what you think Kubernetes is. We’re trying to provide a level playing field. Right now, they are part of the org.
    • SIG Architecture [Jaice Singer DuMars] (confirmed)
      • What we do
        • Manage and maintain architectural consistency over time
        • Manage subprojects:
        • Manage policy and governance
          • API governance (guidance docs and the review process)
          • Deprecation policy
          • Code organization
          • KEP process
          • General issues around Kubernetes scope
      • What are we working on right now?
    • SIG API Machinery [Daniel Smith] (confirmed)
      • Dry run in alpha to see what the predicted outcomes of an action will be
        • You can test API to see what it looks like when it runs through the webhooks
      • CRD versioning change
        • No schema change allowed
        • Register a webhook to do a schema change
        • All the various API definitions should have the same feature set
      • SSH tunnels are going away - this has been deprecated for a year
      • Re: server side apply
        • In a feature branch
          • Allows writing code during freeze
        • More complete designs on the way
      • “We don’t own your API” - we’re not the API reviewers
        • We do own some APIs like metadata format, CRD API, webhook interface APIs, comms between aggregator & APIs, controllers, RBAC API, controller shell (informer, reflector, shared informers, etc.), controller manager binary
      • Upcoming:
        • rate limits, flow control - prevent API quota over-consumption ~ de facto prioritization of API requests
        • Internal API Server coordination - e.g. how do you know when every API is serving the same version of a CRD
        • SIG meeting & agenda ~ if it’s empty the day before, the meeting will be cancelled (there’s an “agenda closed” meeting event)
  • [ 0:00 ] Announcements
    • Calendar information - subscribe, don’t copy! [Jonas]

>>>>> gd2md-html alert: inline image link here (to images/Kubernetes-Community0.png). Store image on your image server and adjust path/filename if necessary.
(Back to top)(Next alert)
>>>>>

alt_text

    *   Google Calendar:

>>>>> gd2md-html alert: inline image link here (to images/Kubernetes-Community1.png). Store image on your image server and adjust path/filename if necessary.
(Back to top)(Next alert)
>>>>>

alt_text

*   Election Update [Jorge]
    *   [All the info you need](https://groups.google.com/forum/#!topic/kubernetes-dev/0gEdp_xdzEI)
    *   Total voters: 677
    *   Actual votes cast thus far: 144
    *   If you are in [voters.md](https://github.com/kubernetes/community/blob/master/events/elections/2018/voters.md) but have not received a ballot please mail [community@kubernetes.io](mailto:community@kubernetes.io)
    *   Election ends on October 3.
  • Kubernetes Contributor Summit is happening [Jorge]
    • Information

    • Content is in draft, SIG leads and TLs, please review and comment

    • Register as part of your Kubecon registration (check the box for collocated events); you will get a follow up email about RSVPing for tracks and

          Sunday dinner and fun @ garage in Seattle 
      
      • Contributor Social Shanghai: 11/13, 6pm, at the convention center
        • Will have panel on contributing from China
  • Contributor Survey [Jorge]
  • [Jorge] October 1st marks the start of Hacktoberfest, a month-long celebration of open source software. This is an opportunity to welcome new Kubernetes contributors to the community. Please help by making an extra effort this month to add more issues with the good-first-issue label. If you have any large tasks that could use help from a lot of contributors, now would be the perfect time to create an issue for it. You can learn more about Hacktoberfest here: https://hacktoberfest.digitalocean.com/
    • ContribEx will be generating a 404 report for new users so they will have a place to go.
    • SIGs, consider updating your good-first-issue labels.
    • SIG Leads, we’ve put together some recommendations for how to give an update for this meeting, the host will be reminding you from now on before your update.
    • Last call for Outreachy intern requests!
  • #Shoutouts!_ (want to say thanks? Use the #shoutouts channel in slack)_
    * @vlad Shlosberg: Huge shoutout to @jorge, @mrbobbytables, @paris, @hubt and a bunch of other for helping make @Foqal a success. Working with me on feedback, helping promote the project, submitting helpful answers, and everything else you guys have done!
    * @bentheelder: Shoutout to @mrhohn for being eternally responsive to networking issues on everything from PR reviews to sig-network test configs, dns images, the network e2es, and answering questions related to network issues in the infra and helping debug! Thanks for helping get the kube-dns manifest images out the door for 1.12! Zihong is always fixing things for us over in #sig-testing :slightly_smiling_face: and now over in #sig-release
    * @nikhita: shout out to @carolynvs for creating lots of help-wanted issues on service catalog (https://github.com/kubernetes-incubator/service-catalog/issues?q=is%3Aissue+is%3Aopen+label%3A%22help+wanted%22)…and having an excellent strategy of doing that! - https://twitter.com/carolynvs/status/1042061098101485580
    * @dims: big shoutout to @fabrizio.pandini for testing all the things! (v1.12 RC1 kubeadm under various scenarios)

#22

September 27, 2018

  • Moderators: Josh Berkus [SIG-Release]
  • Note Taker: Tim Pepper [VMware/SIG-Release]
  • [ 0:00 ]** Release Updates**
    • Current Release Development Cycle [Tim Pepper ~ Release Lead]
      • Release today! …expect published artifacts by ~5pm Pacific
      • Pengfei Ni / @feisky (slack) / @feisker (github) is 1.12.y patch manager.
      • Any issues need targeted for cherry pick to 1.12.1 asap.
      • Retro next week. Enter any notes or thoughts or desires relative to release process in the 1.12 retro doc.
    • Patch Release Updates
      • No 1.10, 1.11 updates
      • 1.9.11 Mehdy Bohlool
        • released today, final update to 1.9.
        • “Finish Him!”
        • Time to upgrade if you haven’t…1.10, 1.11, 1.12 will now be our active releases.
  • [ 0:00 ] **Graph o’ the Week **[Aaron Crickenberger] How to analyze your test failures…
  • [ 0:00 ] SIG Updates
    • SIG Leads, check out this set of recommended topics to cover during your update
    • Please also check the SIG Update Schedule!
      • SIG leadership should prepare ahead of time
      • Give notice to the meeting moderator if your SIG will be unavailable
    • SIG Azure [Stephen Augustus] (confirmed) - Sorry peeps, last minute conflict today that I can’t shuffle
    • SIG Big Data [Yinan Li] (confirmed)
      • Slides
      • Apache Spark 2.4 Release coming mid to late October with Kubernetes functionality:
        • python and R language support
        • Client mode for spark-shell and notebooks
        • Mounting volumes, emptyDri, hostPath, PVC
        • Executor memory request control improvements (fractional values, milli-CPUs)
        • Scheduler backend robustness improvements
      • Future work:
        • Pod templates for driver and executor pod customizations. There are so many options configurable…explosion. Rather than adding more, they’re adding template support to make usage easier.
        • Service shuffling
        • Kerberos authentication
        • Improving support for local application dependencies on the client machines doing submissions
        • Driver resilience, checkpoint/restart for streaming applications.
      • Non-Spark, other projects:
        • Airflow 1.10: new operator and executor for arbitrary pods to run tasks. Operator for life cycle management (newly open sourced)
        • Spark Operator: exporting jvm/driver/executor metrics to Prometheus
    • SIG Scalability [Shyam Jeedigunta] (tentative)
      • Governance: charter is in place, giving official guidance on what the SIG can and can’t do (eg: in the past there’s been a lot of question around what is and is not release blocking)
      • Test resources: github issue (link?) exists to track work items ahead of shifting underlying test resources to CNCF
      • SLOs: API call latency and pod startup latency have been the only two for quite some time. Looking to add multiple new ones for networking.
      • Cluster loader (link): have achieved minimal viable product (MVP) level of readiness. This should enable other teams to easily run scale tests themselves.
      • CI health / tests: Working to speed up pre-submits’ running time, as these have historically been the slowest and are a merge bottleneck (eg: kubemark 500 is waay faster). Investing in de-flaking scale jobs like the 5000 node test, which recently had 8 green runs in a row (a record?).
      • 1.12 release: as typical, close to the end of the release some scalability issues came up, this time in:
        • taint node by condition feature which lead to notably slower start up.
        • CoreDNS regression was masked by other regressions until very late in the release cycle, leaving insufficient time to debug safely and feature dropped (CoreDNS is not the default on GCE yet…1.13?)
  • [ 0:32 ] :mega:Announcements :mega:
    • :postal_horn:Steering Committee Election Announcement :postal_horn:
    • https://www.surveymonkey.com/r/k8s-contributor-2018 -> Contributor Experience survey that will shape our direction and close feedback loops. K8s water bottle for your time today!
    • SIG Chairs/Leads/TLs/Keepers of Zoom and YouTube - check your email for issues relating to those two services. Approx. 40% of our community are not using the correct zoom license. Many, many meetings missing from YouTube channel, which is a negative on transparency.
    • Contributor Summit (Kubecon / CloudNativeCon Seattle) - registration will say waitlist, but: We are not sold out, yet. Please sign up anyway. We are working through a registration issue. There is a possibility we will pull registration from the KubeCon site into a separate registration process. If you have signed up already, you will not need to sign up again. If you are waitlisted, you will be contacted to sort out this issue. Join #contributor-summit on slack for real time info and GH LINK for published information. Initial content will be listed next week.
    • :clap:Shoutouts this week (Check in #shoutouts on slack) :clap:
      • @dims: Big shoutout to @jonasrosland to getting the CNCF meetup off the ground in boston and @abe for his talk on The Kubernetes Release Cycle ( https://www.meetup.com/Cloud-Native-Computing-Boston/ )
      • @timothysc: Huge shoutout to the whole release team and everyone else whose put in crazy effort to make RC2!
      • @aish: HUGE shoutout to @justaugustus for helping us fill the [release 1.13] roster out !
      • @justaugustus Shoutout to everyone who volunteered for the 1.13 Release Team! We staffed a FULL roster of leads and shadows in 16 days and they’ll be in super capable hands with @AishSundar and @spiffxp at the helm!
      • @spiffxp: shoutout to @cjwagner (Cole Wagner) for all of the work he’s done over the past year to remove mungegithub from the project and bring tide to kubernetes/kubernetes

#23

Release Retrospective for 1.12

This community meeting also includes the restrospective for 1.12

October 4, 2018 -


#24

October 11, 2018


#25

October 18, 2018


#26

October 25, 2018

  • :microphone:Moderator: Jorge Castro [SIG Contributor Experience]
  • :memo:Note Taker: Josh Berkus [Red Hat/SIG Contributor Experience/Release]
  • [ 0:00 ]** Release Updates**
  • [ 0:00 ]** Demo **-- Cluster API AWS Provider (chuck@heptio.com)
    • Link to repo
    • Link to slides or docs or whatever goes here.
    • Demo of using the Cluster API to provision AWS.
    • CLI tool: clusterawsadm
      • Creates IAM rules, etc.
    • Must already have SSH key pair (does not create)
    • Starts with an existing Kubernetes cluster to create more clusters; you have to create a 1.11 or later cluster on your own (could be minikube)
    • Create manifests for the CRDs, using “makemanifest”
    • Clusterctl crd then controls the cluster.
      • Pass many parameters by switch
      • Once the new cluster is created, moves the ClusterAPI to that cluster.
    • Secrets? In the CRD defintions.
    • Config file for ClusterCTL? Not sure.
  • [ 0:10 ] :bar_chart:Contributor Tip of the Week [Aaron Crickenberger] :bar_chart:
    • HODL
    • https://prow.k8s.io/command-help#hold
      • Prevents merging
    • http://go.k8s.io/github-labels#do-not-merge/hold
    • /hold to add, /hold cancel to remove
    • Good idea / Bad idea
      • Good idea: explaining why you’re putting on the hold
      • Bad idea: removing a hold in a PR you’re not involved in
    • Reasons to hold
      • Hang on, I the reviewer, think this needs more discussion
      • I, the author, am holding this and will remove it when I’ve heard from the people I want
      • I think the author should have final say on when this PR merges
    • Notes:
      • Anybody can add or remove a hold (don’t even need to be an org member)
      • Can we blacklist? We can, from the org
        • We’d have to restrict hold to org members if it was a problem
  • [ 0:20 ] :satellite:SIG Updates:satellite:
    • SIG Leads, check out this set of recommended topics to cover during your update

    • Slide template if you need it

    • Please also check the SIG Update Schedule!

    • SIG Docs [Jennifer Rondeau, Zach Corleissen] (confirmed)

      • Slides
      • Thanks to zparnold for adding automation foo tracking doc submistions in the Github API
      • Reorganized localization, now under a consolidated repo
        • Everything in kubenetes.io
        • Thanks to Korean translators for making this happen
        • Also updated guidelines
      • Better automation for API reference docs (thanks Chi Ming Tang(sp?))
      • Jennifer Rondeau is new SIG-Docs co-chair
    • Upcoming Work:

      Upcoming doc sprints:

    • Shanghai (localization workflows)

    • Seattle (TBD)

  • Next: how do we ensure that content remains fresh?
  • Localization subprojects have been consolidated into k/website
  • Now have a WG for SIG-Docs tooling, led by Luc Perkins(sp?)
  • Want to contribute? We always need technical reviewers!
    • Or just pick an open issue
    • PRs get more attention than issues, so if you find something wrong, PR a correction.
  • Chairs: Andrew Chen, Zach Corliessen, Jennifer Rondeau
  • SIG Storage [Saad Ali] (confirmed) :satellite:
  • Slides
  • Last Quarter:
    • Topology Aware Volume Scheduling
      • Make scheduler smarter about where storage is
      • Used to be a per-storage hack, now an expressible constraint for the scheduler
      • Started in 1.10, added CSI support in 1.12, beta soon
    • Snapshot & Restore
      • Started a year ago. Was a question whether it should be part of the API at all. But many DB admins would like it.
      • Mapping declarative to imperative was hard.
    • CSI to GA this quarter
      • PV support is primary
      • Also want to support ephemeral volumes, and block volumes
      • Now, we need to migrate the in-tree storage to CSI plugins. This is a blocker for Cloud Provider migration.
        • Challenge: end-users need to have a smooth transition
      • Working on reusable libraries for common storage (iSCSI etc.) that can be used as templates
      • Adding conformance testing for CSI
      • GA depends on completing Kubelet registration mechanism
      • We’re extrating the mount library in k/k to a separate repo so that CSI driver authors can use it.
    • To beta this quarter:
      • Ephemeral volumes
      • CSI Topology
      • CRD automated installation
      • In-tree Block Volume support
  • Catch up with SIG-Storage at their biweekly meeting, or at Kubecon
    • Sessions in Seattle, also a “Cloud-native Storage Day”
  • As part of moving to GA, they need a more robust mechanism than user-modifiable CRDs. THis includes addressing the downgrade problem.
  • [ 0:00 ] :mega:Announcements :mega:
    • Meet Our Contributors - Nov 7th at 230p and 8p UTC

      • 230pm UTC - 5 Steering Committee Members AMA
      • 8pm UTC - contributor mentors AMA
      • #meet-our-contributors on slack
      • YouTube Playlist
    • The final call! CNCF awards nominations are open, details here!

    • Kubernetes Contributor Summit Details

      • The Contributor Social for Kubecon Shanghai has been scheduled. It will be from 5pm to 7pm, November 13, at the convention center. The event will feature a panel of Chinese contributors to Kubernetes, discussing obstacles and opportunities.
      • Seattle - Registration is closed, waitlist is in effect, if you cannot attend please let #contributor-summit (Paris/Jorge/Bob) know so we can free up your slot!

      **:clap: **Shoutouts this week (Check in #shoutouts on slack) :clap:

      • pwittrock - Shoutout to @alexismp @jeefy and @mrbobbytables for helping me with my All Things Open Kubebuilder workshop. Thank you so much!
      • AishSundar - shoutout to @nikopen for automating the Issues and PR spreadsheet for Bug Triage and CI Signal for 1.13 ! Thanks for staying on top of this and accomodating the feature requests
      • jberkus - to @justinsb for splitting out our long-running upgrade tests so that they actually complete
      • spiffxp - Congrats to @bentheelder for creating a PR that deletes over 3 million lines of code https://github.com/kubernetes-sigs/kustomize/pull/503
      • nabrahams - Shoutout to @idealhack for translating a huge pile of slides in preparation for the New Contributor Workshop in Shanghai.
      • kacole2 - Shoutout to @AishSundar @spiffxp @claurence @gsaenger and @ameukam for their help on getting all the k/features (enhancements) issues in a great spot where everything is now being tracked to a PR in k/k and getting the freeze over the finish line.
      • AishSundar - @kacole2 right back at you ! Awesome job on doing all the heavy lifting yourself. Staying on top of ~50 incoming enhancements, following up to prune the list and mentoring the shadows at the same time is no easy feat :slightly_smiling_face:
      • nikhita - Shoutout to @lukaszgryglicki for being extremely responsive to feature requests for DevStats and implementing them and fixing bugs reallyyyyyyy fast!! :tada:
    • Stackoverflow Top Users (Once a month at the end of the month)

    • Community meeting Nov 22

      • We traditionally cancel this meeting due to US Holiday (Thanksgiving)
      • Let’s try to be more global, see #sig-contribex if you want to help drive this meeting this week while the US is out.

#27

Looks like I missed this one, sorry for the delay!

November 1 , 2018 - (recording)

  • Moderators: Tim Pepper [SIG Release / Contrib Ex]
  • Note Taker: Solly Ross
  • [ 0:00 ]** Demo **-- Automation Broker - Michael Hrivnak (mhrivnak@redhat.com) (confirmed)
    • Website
    • Service Bundle: container image/pod that runs to completion to install a service on the cluster
      • Hooks into service catalog via automation broker
      • Ansible Playbook Bundle
        • Easy way to make a service bundle
        • Each service catalog action maps to an Ansible playbook in the bundle
      • Can run other things besides ansible in service bundles (demo on youtube running Helm)
      • Service catalog UI support
        • Partial support in Kubeapps
        • Support in OpenShift
    • Ansible Operator
      • Runs ansible roles/playbooks as an operator
    • Ansible roles exist for manipulating kubernetes objects
  • [ 0:12 ]** Release Updates**
    • Current Release Development Cycle [Aish Sundar - Release Manager]
      • v1.13-alpha3 was cut yesterday, 10/31
      • v1.13-beta0 and Release branch creation scheduled for Tuesday, 11/6
        • Highly dependant on clean CI Signal
        • Branch fast forwards will happen everyday thereafter
      • Code slush is coming up Friday, 11/9
        • Enhancement owners evaluate enhancement readiness based on pending work (code, test and docs)
        • Code Freeze is just 2 weeks away !
        • If you need to enhancement adjusted, please work with the Release team
        • Ensure 1.13 PRs are uptodate on labels (sig, kind, priority, milestone)
        • _Tide will start enforcing Code slush merge label requirements _
      • CI Signal
    • Patch Release Updates
      • 1.11.4 went out last week
      • 1.12.2 went out last week
    • Questions
  • [ 0:00 ] Open KEPs - link to Caleb’s announcement…repository is moving location
    • SIG Architecture is working to improve KEP process for community
    • Extracting KEPs from community repo (see link to the discussion on kubernetes-dev)
    • Try to have small merges that document consensus rather than waiting for full approval/finalization to merge
    • Moving towards eventually making KEPs the main way to propose features (as opposed to being optional)
  • [ 0:00 ] SIG Updates
    • SIG Leads, check out this set of recommended topics to cover during your update
    • Slide template if you need it
    • Please also check the SIG Update Schedule!
    • SIG AWS [Nishi Davidson] (confirmed)
      • https://drive.google.com/file/d/1jDR1Esdu2ApnuLrzsGxn7iv1cU3sWc7R/view
      • We currently host 5 subprojects in SIG-AWS
      • Subprojects aws-alb-ingress-controller, aws-ebs-csi-driver and out-of-tree ccm will be alpha in k8s v1.13
      • Cloud Provider status
        • In-tree
          • Adding e2e tests
          • Will be maintained until out-of-tree is GA, 2 release deprecation
        • Out-of-tree
          • GA Q3 2019
      • CI Signal
        • Added aws-tester plugin
        • Creates ephemeral EKS cluster to run Kubernetes e2e tests as periodic jobs (not blocking)
        • Hoping to integrate etcd conformance tests, cluster API tests as well
    • SIG Scheduling [Bobby Salamat] (confirmed)
      • 1.12
        • Scheduler perf improvements
          • Only score percentage (50%, but configurable) of feasible nodes per pod, properly considered across failure domains
          • Improved affinity/anti-affinity performance
        • Graduated TainNodesByCondiion to beta, which creates taints for node conditions automatically
        • Enable ImageLocalityFunction by default, which prefer nodes which already have the images for a pod, weight set to avoid putting all pods from an RS on the same node
        • Scheduling framework design finalized (move scheduler features towards plugins, both in-process and out-of-process)
      • 1.13
        • Finalize design of gang/co-scheduling (more efficient batch job scheduling, e.g. for ML workloads)
        • Finalize pod scheduling policies (allowing admins to control how pods get scheduled – e.g. preventing setting tolerations, preventing certain namespaces from getting placed on certain nodes)
        • Deprecating the “critical pod” annotation, in favor of pod priority and preemption
        • Enable pod resource limit function (prefer nodes that can fit both a pod’s request and limit)
        • Implement extension points for scheduling framework (see above)
        • Improve equivalence cache (new design to address existing shortcomings)
    • SIG Contributor Experience [Paris Pittman] (confirmed)
      • Update Deck
      • What was done last cycle
        • Theme: making your life easier (automation, documentation, mentoring, events, etc)
        • Performed the contributor survey (graphs on the way!)
          • Common comments
            • Meetups are out of scope, but will pass information on to CNCF
            • You can apply “good first issue” labels even if you didn’t file the issue
            • People liked slack, release team notes in community meetings
          • Scrubbed data is in the link, take a look
        • Misc
          • Communication moderation changes (stay public while dealing with bad actors) – SIG chairs should learn how to follow these processes
            • Calendar is private ATM because of bad actors
            • Zoom links aren’t publicly posted for similar reasons (please don’t tweet them), but work is being done to solve this with Zoom
          • Launched discuss.k8s.io as a community forum, please post/take a look!
      • Upcoming
        • Revamp developer guide
        • Move KEPs out kubernetes/kubernetes
        • Build a contributor site
        • Upgraded communications guide
        • Improve SIG Chair processes (e.g. Zoom-to-Youtube automation process) – please reach out if you have opinions
      • Seattle Contributor Workshop
        • Waitlisted (if you’re a chair, TL, or subproject owner who hasn’t signed up, please reach out!)
        • lots of good content planned
        • Night-before event to hang out and talk
      • Consider mentoring, even if it’s just 1 hour per quarter
        • Only need one merged PR to be a mentor
      • See slides for a _whole lot _more work, information, links, and sigup information
  • [ 0:00 ] :mega:Announcements :mega:
  • Shoutouts
    • Nikhita: shoutout to @dims for being Asia/EU friendly while deciding the meeting time for #k8s-infra-team
    • Mzee1000: Shout-out to @mrbobbytables for his help with Kubernetes 101 in Bangalore
    • Jberkus: to: @justinsb for continuing to be the “difficult test fail” resolver.
    • Jberkus: to @neolit123 for fast turnaround on kubeadm test fails
    • Fejta: shoutout to @bentheelder for finally creating a @thockin emoji :thockin:
    • @liz to: @bentheelder for going above and beyond to help me get my KIND tests working!
    • @paris thanks to @nikhita @roycaihw @brendanburns @dims and many others for answering questions from first time contributors in the outreachy process slack channel #outreachy-apps
    • @spiffxp thanks to @audreylim for tackling e2e test error messages as her first kubernetes pull-request (https://github.com/kubernetes/kubernetes/pull/69583)
    • To Solly Ross for taking notes today

#28

November 8, 2018 -

  • Moderators: Jorge Castro [SIG Contributor Experience]
  • Note Taker: Solly Ross (Google/SIG AUtoscaling)
  • [ 0:00 ]** Demo **–IngressRoute with Contour - Steve Sloka (steves@heptio.com)
  • [ 0:00 ]** Release Updates**
    • Current Release Development Cycle [Aish Sundar - Release Manager]
      • We cut Beta0 and the 1.13 Release branch yesterday, 11/7 !
      • Updated to Go 1.11.2 before Beta.
      • Code Slush is this Friday (tomorrow), 11/9
        • All PRs need /priority, /kind, /sig, /milestone labels to merge post 5pm PST
      • Code Freeze is just a week away - 11/16
        • Handful of Enhancements are pending only tests and docs
        • There are a couple of Enhancements at risk and a few that have had no activity in past week or so.
        • Owners please update k/enhancement issues with current status
        • Please reach out to Release team early on if you think you need to move out your enhancement
      • CI Signal
      • Docs
        • Open PRs: 11/22
        • Completed PRs: 2/22
        • We have 7 outstanding PRs. We will be pinging owners on issues.
      • Release Notes
        • Sig Leads expect initial draft of the release notes coming your way for review next Monday 11/12
        • Please leave early feedback if you can
      • Questions:
        • Where should we send 1.14 volunteers
          • Look out for issue for more info, will be linked in next week’s update
    • Patch Release Updates
      • x.x
      • y.x
  • [ 0:00 ] **Contributor Tip of the Week **[Aaron Crickenberger]
  • [ 0:00 ] SIG Updates
    • SIG Leads, check out this set of recommended topics to cover during your update
    • Slide template if you need it
    • Please also check the SIG Update Schedule!
    • SIG Cluster Lifecycle [Tim St. Clair] (confirmed)
      • Slides
      • Homepage: https://contributor.kubernetes.io/sigs/sig-cluster-lifecycle/
      • Mission: simplify creation/upgrade/downgrade/teardown of Kubernetes clusters and their components
      • Last cycle
        • Kubeadm: Config changes, improved CRI, HA, cert management, air-gapped support
        • ClusterAPI: Provider-specific repos, many providers
      • Upcoming plans
        • Better meeting times and subproject coordination
        • Kubeadm to GA (beta config, command line options fully supported)
        • Cluster API integrating cluster API into kops
        • Kubespray defaults to kubeadm
        • ComponentConfig for structured configuration of other Kubernetes components
        • Deprecate: kube-up, kubernetes-anywhere
      • Reminder: for upgrade testing: SIG cluster lifecycle owns the framework, others own the actual tests
      • Events: Upcoming kubecon talk on the future of addons
      • New etcd management proposal/tooling (proposal just approved)
      • Questions:
        • If kube-up and kubernetes-anywhere are deprecated, what’s the standard deployer (what’s the kubeadm version of curl | bash to install)?
          • documented well on the kubeadm section of the docs
          • kubernetes-anywhere was just used for e2e tests
          • other tooling (e.g. kubespray) builds on top of/orchestrates kubeadm, kubeadm only sees local machine
          • Cluster API provides the same view of different backends, kubeadm is the base layer, providers in the middle
        • Why use CRDs instead of aggregated APIs?
          • maturity, easy of use, portability
    • SIG OpenStack [Chris Hoge] (confirmed)
      • Slides
      • Previous work
        • In-tree driver is deprecated and will go away soon
        • Manilla provision
        • CSI support for Manilla and Cinder
        • Magnum is now Kubernetes Certified Installer
        • Driver for Cluster API in the works
      • Future Work
        • Heat & Senlin autoscaling drivers
        • Storage driver consolidation
        • Barbican driver for key management
        • Finish in-tree code removal
      • Transitioning into a WG under SIG Cloud Provider
      • Events:
        • OpenStack Summit, Berlin (Nov 13-15)
        • Sessions at KubeCon Seattle
    • SIG Auth [Mo Khan ] (confirmed)
      • Slides
      • Homepage: https://contributor.kubernetes.io/sigs/sig-cluster-auth/
      • Features
        • Per-pod ephemeral service account tokens (projected volumes instead of secrets)
          • If NOT using client-go today, need to keep reading token off disk
        • Restricting Kubelet self-applied labels (via an admission plugin)
        • Dynamic audit configuration (add/remove audit sinks without restart of API server)
      • Container Identity WG winding down
  • [ 0:00 ] :mega:Announcements :mega:
    • Contributor Summit @ Kubecon

      • Shanghai: Josh is getting on a plane, see you all there!
      • Seattle: Chairs and owners, if you haven’t confirmed we’re running out of time, please let us know.
    • Community Meeting Schedule

      • 11/22 (Thanksgiving in the US) - Meeting is Still on, Ihor will be your host!
      • 12/6 - Release Retro for 1.13 (tentative!)
      • 12/13 - Kubecon, no community meeting
      • 12/20 and 12/27 - No community meetings
      • January, SIG Apps, SIG UI, SIG VMWare
    • Steering committee not having meeting in two weeks, will have one just before Kubecon

      • SIGs: Please try to have your charters in by KubeCon

      **:clap: **Shoutouts this week (Check in #shoutouts on slack) :clap:

    • paris and jdumars: big thanks to @mattfarina who just spent an hour helping organize our project boards

    • Jberkus: Shanghai Shoutouts for next week: Megan Lehn for doing all the logistics and legwork from thousands of km away, @puja @xiangpengzhao and @idealhack for translating all the New Contributor Summit materials and many other things besides! Also, to @mrbobbytables and our localization volunteers for getting the international forums at dicuss.kubernetes.io launched!

    • AishSundar: Shoutout to @jberkus and his team of CI Signal shadows @maria and @mortent for staying on top of CI signal failures and flakes every day, opening and following up on test issues and fixes and help maintain stable test health for 1.13 release !

    • jdumars - Big thanks to @spiffxp @dims and @mattfarina — all of whom have stepped up and helped with the work in SIG Architecture!

    • mauilion - shoutout to @jdetiber for always finding time to help dig into the cluster-api stuffs.

    • AishSundar - shoutout to @justinsb, yet again, for extremely quick turnaround on a long standing Upgrade testing issue (#56787). This helped us get clean e2e CI coverage one of the 1.13 Beta Feature “Taint Based Evcitions”


#29

November 15, 2018

  • Moderators: Jorge Castro [SIG Contribex]
  • Note Taker: Solly Ross (SIG Autoscaling/Google)
  • [ 0:00 ]** Demo **–Pulumi - an OSS, k8s-native deployment orchestration engine [Alex Clemmer]
    • Link to slides
    • Link to repo
    • Pulumi: open-source tools for managing cloud infrastructure
      • Declare steady state (like Kubernetes) using programming languages like Python, Javascript, Typescript to manage the cloud repos
      • Declare desired “outputs” to be saved for easy access (e.g. Service IPs)
      • Schema is _exactly _kubernetes schema for Kubernetes types, etc (no special other format)
      • Knows how interact with deployments (has concept of updates, knows that it needs to rollout, wait for rollout to succeed, only delete old objects after)
    • Workflow for using something like RDS:
      • Without Pulumi: deploy using one tool (e.g. terraform), then fetch connection string into secret (maybe using something else), then use in kubernetes (e.g. deploy app via Helm)
      • With Pulumi: Declare steady state in code for everything
    • Live demo: deploy CosmosDB + Helm chart
      • Declare CosmosDB, exported connection string
      • Declare secret (using normal Kubernetes schema) with connection string
      • Declare Helm chart (deploying Bitnami Node.JS image) using secret to supply external DB
      • How it works:
        • Pulumi figures out dependencies automatically to figure out that CosmosDB needs to come before Secret, chart depends on secret
        • pulumi up will show a “plan” of operations + Kubernetes JSON, executes plan on confirmation
      • Can specify “stack outputs” to save from the objects generated, to fetch programmatically (e.g. IP of serving generated by Helm chart)
    • Question
      • How is schema generated, what happens to unknown attrs
        • Schema is generated via OpenAPI spec based on all available versions
  • [ 0:00 ]** Release Updates**
  • Patch Release Updates
    • v1.12.3 cut planned Monday, Nov 26th
    • V1.10.0 published earlier this week
  • [ 0:00 ] **Contributor Tip of the Week **[Jeffrey Sica]
    • cs.k8s.io – Search all repos in seconds
      • Can regex search across all Kubernetes repos and orgs
      • Automatically filters out certain types of files, but that can be configured
    • Check out the contributor cheatsheet for other shortcuts (PRs accepted!)
  • [ 0:00 ] SIG Updates
  • [ 0:00 ] :mega:Announcements :mega:
    • Contributor Summit @ Kubecon

      • Shanghai: Great turn out! Lots of great pics on Twitter etc.
      • Seattle: Chairs and owners, if you haven’t confirmed we’re running out of time, please let us know. You do not need a ticket to kubecon/cnc for this. Email community@kubernetes.io
    • Kubecon US is SOLD OUT. If you register now you’ll be waitlisted. \

    • Community Meeting Schedule - there are no SIG updates for December.

      • 11/22 (Thanksgiving in the US) - Meeting is Still on, Ihor will be your host!
      • 12/6 - Release Retro for 1.13 (tentative!)
      • 12/13 - Kubecon, no community meeting
      • 12/20 and 12/27 - No community meetings
      • January: SIG Apps, SIG UI, SIG VMWare \
    • Steering committee not having meeting in one weeks, will have one just before Kubecon

      • SIGs: Please try to have your charters in by KubeCon
        • Each SIG has been given a steering committee member to review
        • reach out to steering committee if you have questions
    • Office Hours next week on YouTube:

      • See this for more information
      • Come answer questions about Kubernetes on a livestream!
  • [ 0:00] :clap: Shoutouts this week (Check in #shoutouts on slack) :clap:
    • paris - very big shoutout to @jberkus and the entire kubecon shanghai new contributor workshop team! josh built a team and carried out the event plan for this first time, sold out event in a new market to welcome contributors from this region. the event is in a few hours (from this timestamp) - best of luck and have a great time team!!
    • jberkus - TY! Let’s add all the names: @tpepper @puja @nabrahams @xiangpengzhao @idealhack & Megan Lehn & Jerry Zhang
    • neolit123 shoutout to @fabrizio.pandini for organizing the transition of phases in kubeadm to GA and also thank you to all the new kubeadm contributors who helped us with this work @yago @yuexiao wang @ereslibre @Rohit
    • spiffxp Shoutouts to @chenopis @zacharysarah and @bradtopol for organizing and running the docs translation sprint at kubecon Shanghai
    • Ivan Font @bentheelder and all others who worked on kind: I wanted to give a shout out for the work done to create kind. Nice work! I’ve experimented to get kind working with multiple clusters so that we can use it to test federation-v2 with multiple clusters for dev and CI and I’m very impressed with it so far! I filed a few issues #110, #111, #112, and #113 that I’ve stumbled across in the process of doing that, but it is not a reflection of the quality of work that’s been done here. Again, thanks for the awesome work! Thanks to @munnerz @neolit123 @Jorgealarcon @Lion-Wei @TaoBeier @amwat

#30

November 29, 2018 - Last meeting of the year!

  • Moderators: Josh Berkus [SIG-Release]
  • Note Taker: Solly Ross [Google/SIG Autoscaling]
  • [ 0:00 ]** Demo **-- Docs Modeling Working Group Demo [Andrew Chen, @chenopsis, Dominik (dominik.tornow@sap.com)] (confirmed)
    • Link to slides
    • Modelling how we design and look at documentation
      • Idea:
        • Ideally, two people who look at the same system develop the same mental model
        • Looking a documentation, there may be encoding/decoding loss (writing/reading docs), which leads to different mental models
      • Fundamental Modeling Concepts
        • Approach to system modeling with formal models of system’s structure and behavior
        • Diagrams and formal models can help show whole-picture view
        • Show people how things work without needing to point people at actual source code
    • Issues with existing docs:
      • docs are task focused (good for on-demand “how do I” type questions), but can’t easily develop a coherent general picture
      • non-obvious behavior doesn’t match general mental model, docs should help fix that
    • Process:
      • Ongoing: Discuss models (in SIG Docs), Interview engineers, validate models, create source materiel (e.g. Medium posts) and get feedback
      • Eventually: fold back into to k8s.io (planned for next year)
  • [ 0:13 ]** Release Updates**
    • Current Release Development Cycle [Aish Sundar - Release Manager]
      • Code freeze for 1.13 is now lifted! Code thaw went into effect 11/28, 8pm PST.
      • Master is now open for 1.14 development.
      • Only the absolute most critically urgent bug fixes might be cherry picked back in time for 1.13.0.
      • 1.13-rc.2 slated to cut tomorrow, 11/30.
      • The release is on target for** Monday, 12/3/2018**, pending CI signal.
      • If you still have outstanding Docs PR or Release notes, please get -them in ASAP.
      • We’re targeting our release retrospective for next week’s Community Meeting on 12/6. Please add any comments you’d like included in discussion for things that worked well and things that should change in our 1.13 retrospective document.
    • Patch Release Updates
      • 1.12.3
      • 1.11.5
      • 1.10.11
  • [ 0:16 ] SIG Updates
  • Please drop a note in the community meetings doc, or reach out if you want to talk about a KEP in the community meetig
  • [ 0:00 ] :mega:Announcements :mega:
    • Contributor Summit [Paris and Jorge]
      • **We are sold out/waitlisted - L A S T C A L L if you’re a SIG Chair, TL, or subproject owner **
      • Talks have been added to the community calendar, shortcut: http://bit.ly/kubernetes-summit
      • Check out #contributor-summit on slack
      • Event information
    • Community Meeting Schedule - there are no SIG updates for December.
      • Today is the last “normal” community meeting
      • 12/6 - Release Retro for 1.13 (tentative!)
      • 12/13 - Kubecon, no community meeting
      • 12/20 and 12/27 - No community meetings
      • January 1/3 : SIG Apps, SIG UI, SIG VMWare
    • Meet Our Contributors will be 5 December.
      • Steering Committee AMA @ 730a PT / 330pm UTC
      • Mentor panel @ 1pm PT / 9pm UTC
      • Be a mentor to hundreds with one hour of your time! Reach out to parispittman@google.com / “paris” on slack to get scheduled.
    • No k8s office hours this month - thanks to all the volunteers who helped make the program a success this year.
    • **:clap: **Shoutouts this week :clap:
      • Twitterverse shoutouts for our fearless 1.13 Release Team Lead, @AishSundar: https://twitter.com/stephenaugustus/status/1063610123149545472?s=19
      • Shoutout to @amerai for adding a search bar to Testgrid so that you don’t have to dig to find the right dashboard! https://testgrid.k8s.io/
      • to @mkimuram & @saad-ali & @msau42 for rapid response to multiple storage test issues with new features.
      • to @mrhohn for fast & insightful help with sig-network test failures
      • Huge shoutouts to the entire 1.13 Release leads and shadows for their stellar efforts at every stage throughout the cycle, enabling us to stabilize and hopefully land the release on time - @kacole2 @jberkus @cjwagner @dougm @nikopen @tfogo @marpaia @kbarnard10 @spiffxp @tpepper@aleksandram!
      • Special shoutout to contributors “technically” not on the release team, but have been instrumental in getting us unblocked at numerous points this release with their reviews, test fixes and test-infra support - @dims @liggitt @justinsb @cblecker @bentheelder @justaugustus (edited)
      • shoutout to you @AishSundar for keeping us all in line this whole cycle! you’ve been a totally awesome release lead.
      • Shoutout to @mrbobbytables for significally reducing my admin overhead for the New Contributor Workshop!