cncf / cluster Goto Github PK
View Code? Open in Web Editor NEW🖥🖥🖥🖥CNCF Community Cluster
Home Page: https://cncf.io/cluster
🖥🖥🖥🖥CNCF Community Cluster
Home Page: https://cncf.io/cluster
If you are interested in filing a request for access to the CNCF CIL, please fill out the details below.
Lauri
Ojansivu
CEO of my one person company
Wekan
Open Source Trello-like kanban
All that want to self-host kanban board and have control of the data.
Website:
https://wekan.github.io
Repo:
https://github.com/wekan/wekan
MIT License:
https://github.com/wekan/wekan/blob/devel/LICENSE
Current platforms where Wekan runs:
https://github.com/wekan/wekan/wiki/Platforms
Servers at Europe, Amsterdam.
Type 3, Type 1E and Type 2A
Other uses:
Internet.
I have 920 contributions to various repos at GitHub in the last year:
https://github.com/xet7
I'm maintainer of Wekan, 444 commits:
https://github.com/wekan/wekan/graphs/contributors
I have helped to keep Wekan alive, by continuing Wekan in a fork, and got a chance to merge it back:
https://github.com/wekan/wekan/wiki/FAQ#what-was-wekan-fork--wefork
Wekan has many integrations coming with IFTTT with Huginn or Flogo, Gogs git repo hosting, etc at Integrations wiki page.
For these, advancement is in having Docker Compose script and other ways to orchestrate on various public and private cloud platforms, easy ways make more integrations, and easy deployment and management of Wekan and other integrated Open Source software.
Wekan wiki press page has link to Hacker News article where someone did say that they use Wekan to store confidential patient data in Wekan, so Wekan is mission critical for some.
I have worked as SysAdmin and AWS Cloud Architect, so most things I could probably be able to setup myself, though help is very welcome, like in setting up Triton, Proxmox etc environments. If you have some example Terraform or Ansible scripts to help with setting up various different environments, it would help.
For Google Cloud, I currently don't know how to enable websockets so Wekan works. I also have not tried Wekan yet at Azure. More about this at Platforms wiki page.
Jeremy
Eder
Red Hat
Engineer
Deploying 1000 nodes of OpenShift on the CNCF Cluster (Part 2)
We are interested in:
Working through the operational concepts necessary to handle a large bare metal scale-out environment.
Comparing the behavior of Kubernetes on OpenStack with Kubernetes on bare metal.
Run our newly developed workload generators and test suite
Utilizing newer features in Kubernetes to make use of bare metal hardware features.
To compliment our earlier work on the CNCF lab (https://cncf.io/news/blogs/2016/08/deploying-1000-nodes-openshift-cncf-cluster-part-1) we would like to propose a full-lab scale test scenario once the CNCF lab is at full capacity. We will look to quantify improved performance when running on bare metal instead of virtualized. We will conduct some specific HTTP load testing and storage (persistent volume) performance testing.
Yes, we will use our pbench framework https://github.com/distributed-system-analysis/pbench to capture metrics on each run. We expect this to involve Prometheus, a CNCF project, to the extent that we use it for gathering Kubernetes API server metrics.
Kubernetes, Prometheus, end users who are looking to run high performance workloads on bare metal environments. Also fluentd if that is accepted (OpenShift uses fluentd for logging).
Yes: https://github.com/openshift
Yes, we have already open-sourced everything we write and we have shared significant amounts of data via blog and public-speaking engagements at industry conferences.
Yes.
Not that we are aware of. We have good experience handling OpenShift at scale and we are proposing a two-phase approach where we prototype on 100 nodes (this proposal) with an adjacently-scheduled phase at full-lab scale of 1000 nodes.
Yes.
Deploying 1000 nodes of OpenShift on the CNCF Cluster (Part 1)
Over 30 bugs were filed across projects such as Kubernetes, OpenShift and Ansible.
No.
Red Hat is a fully open-source company. Red Hat is a platinum founding member of CNCF, a contributor to docker, kubernetes, openshift origin, and many more.
Phase 1 (this request) 100 compute nodes and 10 storage nodes.
Phase 2 (upcoming request) 1000 compute nodes.
Phase 1 two weeks. Phase 2 two weeks. Both phases must be adjacent in order for us to get what we need done in the 2 week periods.
With, RHEL7.3
We are working to push beyond control plane scalability to simulate realistic bare metal scenarios. This will include loading applications that represent an accurate mix of what we have seen in the wild. Being able to do this at higher scale levels will help us to discover best practices from an architecture standpoint as well as to help validate capacity planning formulas to see if they hold up at higher scale and load levels.
If you are interested in filing a request for access to the CNCF Community Cluster, please fill out the details below.
If you are just filing an issue, ignore/delete those fields and file your issue.
Qi
Liu
PingCAP, Inc.
CEO
TiDB
TiDB is targeting an easier deployment to provide a scalable and consistent database on Kubernetes and other container environments.
TiDB is a distributed SQL database and it supports the best features of both traditional RDBMS and NoSQL.
We will integrate the Prometheus metrics to measure the performance to improve the reliability, scalability and performance. We are hoping to get the load conditions on every node such as the IO , CPU, etc.
etcd, Kubernetes, and Prometheus.
Yes. https://github.com/pingcap/tidb
Yes and Yes.
Yes. From the very beginning, the deployment and testing of TiDB depends on the containers. Currently, all the test cases and continuous integration are based on Docker.
No.
No.
The co-author of the following open source projects:
TiDB: https://github.com/pingcap/tidb
TiKV:https://github.com/pingcap/tikv
Codis: https://github.com/CodisLabs/codis
100 nodes.
2 weeks.
With. Ubuntu 14.04+ or CentOS 7. Ubuntu 14.04+ is preferred.
As a scalable and consistent database, TiDB deployment supports the applications and microservices on the cloud native platforms so that they don’t have to care about the state and will be very easy to scale. This testing could help us ensure that.
We need to use Kubernetes and the latest Docker.
If you are interested in filing a request for access to the CNCF CIL, please fill out the details below.
If you are just filing an issue, ignore/delete those fields and file your issue.
Thomas
Felder
N/A
dev
analyticssandbox
data-mining / social sentiment analysis experiements / container orchestration etc
data scientists
yes – https://github.com/tfmolch
3 midsize VMs
debian
none worth mentioning yet – I'm a self-taught programmer!
well at the very least I'll be testing/bug reporting
SAN storage shared between the nodes would be a plus!
Currently the template links users to find out about supported operating systems on Packet's /bare-metal/ page. We should consider updating to a more directly informative page:
https://help.packet.net/technical/infrastructure/supported-operating-systems
Stuart
Williams
Weaveworks
Director of Product Management
Zero-downtime Upgrade Research
Infrastructure components are now often containerized, upgrading them without causing, or minimising downtime presents challenges that are not adequately explored in cloud native environments.
We will try various scenarios, at scale, that upgrade container-based infrastructure and validate that applications using those services are impacted or impeded in their operation.
Liveness of applications and error detection
Users of containerized infrastructure components will benefit from our learnings.
Yes. Results will be published & changes to OSS described in a blog
No
No
n/a
n/a
n/a
All Weave products are OSS; our employees have created and contributed to many OSS projects - from Linux kernel patches, to ASF & Eclipse projects, to founding RabbitMQ.
1000 nodes
120 hours
With OS.
Not yet.
If you are interested in filing a request for access to the CNCF Community Cluster, please fill out the details below.
If you are just filing an issue, ignore/delete those fields and file your issue.
David
Aronchick
Product Manager
Kubernetes
Testing scale on bare metal
We want to test scale on bare metal.
Yes, whether or not we pass kubemark.
All
Yes, we will change benchmarks.
No
no
10000 Nodes
72 hours
With
No
Alan Gutierrez ~ @bigeasy ~ hacker
Cory Kaufman-Schofield ~ @allspiritseve ~ also hacker
CNCF Demo on ARM ~ Get the CNCF demo applications working on ARM. Initiate a project to get the CNCF stable of hosted projects working on ARM using the CNCF demo application as a test bench. Begin by first getting the demo to work with the Packet infrastructure. This will create a patch to the demo where the AWS installation step can be replaced by a deployment Packet x86. Once working on x86 we can then begin to port the demo components to ARM on Packet peicemeal with integration testing aginst that x86 hosted components. The end result is a CNCF demo that can be hosted mix or match on x86 or ARM at Packet, verified function of the CNCF components on ARM, along with any insights gathered in porting CNCF's Go based components form x86 to ARM.
Anyone working with cloud native applications. The heart of CNCF is Kubernetes, so this will be a demonstration of the ability to host a constainer orchestration environment on ARM hardware. Specifically this will be an exercise of Kubernetes container orchestration and Prometheus container monitoring.
Yes, the code will be 100% open source and submitted back into the CNCF demo.
The project will need access to 3 machines for the x86 portion of the port, each with a minimum of 32 GB RAM. The Packet Type 1 server should be adequate to the task. Type 1 servers are hosted in data centers with the Packet Block Storage service so it would allow us to evaulate the use Kubernetes block storage plugin for Packet Block Storage. Upon completion of the x86 evaulation, a similar number of Type 2A servers will be required for the Arm portion of the effort.
Timothy
St. Clair
Red Hat
Principal Software Engineer
Kubernetes Scalability SIG
Compare and diagnose scale test results across companies.
Scaling Kubernetes (Scalability SIG)
Yes
The scale sig
Yes
Not yet.
no
kubernetes maintainer
Active SIG lead and maintainer
n/a
I'm an active kubernetes maintainer.
~300
4 weeks
Without, or RHEL
Ideally I think openstack-ifying would be beneficial to the community.
If you are interested in filing a request for access to the CNCF Community Cluster, please fill out the details below.
If you are just filing an issue, ignore/delete those fields and file your issue.
Justin
Garrison
Self
Reference architecture
Cloud Native Infrastructure book
Resources needed to test and set up examples of CNCF projects to verify functionallity and infrastructure topologies.
@kris-nova and I are working on a book for O'Reilly about cloud native infrastructure. We need some resources to set up and test CNCF projects so we can create diagrams and verify functionallity of the projects and explain why specific design patterns are beneficial.
We'll be setting up various projects (kubernetes, fluentd, linkerd, and prometheus) and need to be able to test the projects function as intended in an environment not directly tied to any cloud provider.
Any of them who read the book (hopefully everyone 😄)
I do not believe the book will be open source although I will ask if it's a possibility. Otherwise the content/results will be available for anyone who is able to buy the book.
Any issues we find during testing we will open cases with the projects to discuss our issues and use cases.
yes, we will be deploying all projects on top of Kubernetes.
There is a risk the infrastructure will take to look to provision on bare metal. I am not sure what options are available for deploying into the cluster.
no
@kris-nova is an active participant in the kubernetes community and maintainer of kops
I am an active user/contributor of kubernetes
30
any
2 weeks (the book is a 4 month project, we can submit multiple issues for more scheduled time)
Ubuntu
Provide reference architecture and examples
Stephen
Day
stephen at docker dot com
Docker, Inc/CNCF/containerd
Maintainer/Software Engineer
containerd
An open and reliable container runtime: https://containerd.io.
At this time, the cluster resources requested will be used to support alternative platform builds, including but not limited to ARM, s390x and power systems. While we would like to use hosted services, the support for less popular architectures.
These resources will be used to build containerd
on every pull request, merge and build releases on alternative platforms. This will allow contributors to have confidence in their changes without having to try them across platforms. This will benefit maintainers by ensuring that testing is equal across platforms. Users will benefit from a better tested container runtime.
We hope to also benefit the kubernetes community by providing a stable container runtime that will work across any platform.
The ARM community is steadily growing and we see stability there as an important part of containerd's offering. s390x and power are heavily used in enterprise applications.
Yes, https://github.com/containerd/containerd
amd64
head node for running builds.
We'll also need a place to back up build database in case of catastrophic failure. We will not require hot standby, as impact of failure will be low.
Ubuntu 17.04
I am a maintainer on Moby, OCI Image Spec, containerd, SwarmKit, Docker Registry and other smaller projects.
With containerd
at the base of containerization, it is a critical component of the cloud native "water cycle". Having solid cross-platform support is a must.
Thanks to all those that have helped with getting this proposal together!
cc @containerd-maintainers
If you are interested in filing a request for access to the CNCF Community Cluster, please fill out the details below.
If you are just filing an issue, ignore/delete those fields and file your issue.
Allan
Naim
Product
Cluster Federation testing
Cluster Federation testing between bare metal CNCF clusters and GKE
Sig-federation
Yes
Yes, containerized workloads running in Kube
n/a
no
n/a
no
Cluster federation Kube upstream
Only need about 4 nodes
n/a
2 wks
Benefits the cluster federation effort within Kube
If you are interested in filing a request for access to the CNCF Community Cluster, please fill out the details below.
If you are just filing an issue, ignore/delete those fields and file your issue.
Chris
Aniszczyk
Linux Foundation / CNCF
VP / COO
CNCF Marketing Demo (using Boinc)
Builds a canonical marketing demo using CNCF technology
Builds a canonical marketing demo using CNCF technology
Application performance
All of the CNCF community looking to learn how to build a non-trivial CNCF application
Everything will be open source from day 1
Nope
No
N/A
N/A
N/A
N/A
20
A couple of weeks to start, may request permanent access
With
We are CNCF staff :)
Jeremy
Eder
Red Hat
Engineer
Deploying 1000 nodes of OpenShift on the CNCF Cluster (Part 2)
We are interested in:
Working through the operational concepts necessary to handle a large bare metal scale-out environment.
Comparing the behavior of Kubernetes on OpenStack with Kubernetes on bare metal.
Run our newly developed workload generators and test suite
Utilizing newer features in Kubernetes to make use of bare metal hardware features.
To compliment our earlier work on the CNCF lab (https://cncf.io/news/blogs/2016/08/deploying-1000-nodes-openshift-cncf-cluster-part-1) we would like to propose a full-lab scale test scenario once the CNCF lab is at full capacity. We will look to quantify improved performance when running on bare metal instead of virtualized. We will conduct some specific HTTP load testing and storage (persistent volume) performance testing.
Yes, we will use our pbench framework https://github.com/distributed-system-analysis/pbench to capture metrics on each run. We expect this to involve Prometheus, a CNCF project, to the extent that we use it for gathering Kubernetes API server metrics.
Kubernetes, Prometheus, end users who are looking to run high performance workloads on bare metal environments. Also fluentd if that is accepted (OpenShift uses fluentd for logging).
Yes: https://github.com/openshift
Yes, we have already open-sourced everything we write and we have shared significant amounts of data via blog and public-speaking engagements at industry conferences.
Yes.
Not that we are aware of. We have good experience handling OpenShift at scale and we are proposing a two-phase approach where we prototype on 100 nodes (this proposal) with an adjacently-scheduled phase at full-lab scale of 1000 nodes.
Yes.
Deploying 1000 nodes of OpenShift on the CNCF Cluster (Part 1)
Over 30 bugs were filed across projects such as Kubernetes, OpenShift and Ansible.
No.
Red Hat is a fully open-source company. Red Hat is a platinum founding member of CNCF, a contributor to docker, kubernetes, openshift origin, and many more.
1000 nodes. (We realize that there will be slightly less than 1000 available for us to use).
2 weeks at least.
Please schedule this immediately after #21 so that we can retain our existing environment, and expand it on to the additional nodes.
With, RHEL7.3
We are working to push beyond control plane scalability to simulate realistic bare metal scenarios. This will include loading applications that represent an accurate mix of what we have seen in the wild. Being able to do this at higher scale levels will help us to discover best practices from an architecture standpoint as well as to help validate capacity planning formulas to see if they hold up at higher scale and load levels.
The cluster is currently unavailable due to a serious issue in the frontend cluster. We are working on making it operational but currently it is not possible for tenants to access the cluster in any way. We will post updates here and on CNCF Cluster Slack channel once we have more information. I am extremely sorry for any trouble it might cause.
Malte
Schwarzkopf
Cambridge Systems at Scale (CamSaS) / Firmament.io / MIT CSAIL
Researcher
Firmament and Poseidon scale-up evaluation
In our research, we developed the Firmament cluster scheduler, on which we have an upcoming paper appearing in OSDI 2016. In this paper, we evaluate the scalability of Firmament’s min-cost flow optimization-based scheduling by replaying workload traces, and in a small-scale test-bed deployment. We would like to perform a real-world scale-up evaluation to further validate our results, and to explore the benefit of Firmament’s scheduling policies for real workloads running on a larger-scale clusters than we currently have access to.
Poseidon (https://github.com/camsas/poseidon) is an integration of Firmament’s scheduling logic with Kubernetes: we have developed Poseidon as a drop-in replacement for the default Kubernetes scheduler (kube-scheduler). We would like to deploy Poseidon at scale and measure its performance in order to compare it against our own stand-alone research cluster manager and ensure that there are no undue performance penalties in either implementation.
Yes, we hope to measure the following:
The Kubernetes project, via the Poseidon scheduler, which we are making available for anyone to use. Poseidon is API-compatible with kube-scheduler
and can be deployed on existing Kubernetes clusters.
Yes, see https://github.com/camsas/firmament and https://github.com/camsas/poseidon.
Yes, we open-source all of our work on Firmament and Poseidon under the Apache 2.0 license. Our experimental results will be published as part of a peer-reviewed paper, a blog post, or a technical report.
Yes, Poseidon schedules containers (via Kubernetes). Firmament’s standalone cluster manager has alpha support for container execution; we hope to improve this in the near future.
The only main risk is that the timing of cluster access may not work out well, e.g., coinciding with a major deadline, teaching, or a conference. We are a small team of two researchers, so our cycles are limited. We may also discover previously unknown scalability bottlenecks as a result of the testing, although we would consider this a success rather than a risk.
No.
Our research software, data sets, experimental infrastructure, and scripts are open-source (https://github.com/camsas), and we contribute to the Kubernetes SIG-scheduling interest group.
If we use the CNCF cluster for a scale-up experiment mentioned in a research publication or blog post, we will acknowledge the support by the CNCF as we would acknowledge any other industrial support of our research.
100, with possibility to increase to 500.
1 week.
With OS, if Ubuntu (14.04 LTS or 16.04 LTS) is available.
It will provide better container scheduling, potentially tighter packing of workloads with reduced interference, and validate the scalability of the min-cost flow optimization scheduling approach in a real-world environment.
Some of our experiments involve running applications that generate a high network load (i.e., saturating a 10 Gbit/s link).
If you are interested in filing a request for access to the CNCF Community Cluster, please fill out the details below.
If you are just filing an issue, ignore/delete those fields and file your issue.
Thanatas
Pongpanotkorn
[email protected];[email protected]
Developer
xaxiclouddev,.inc.co.th
Oomnoi Krathumban Thailand 74130
Auto deployer
No, i don't have knowhow.
Yes
Undentified it.
ZiplineStatusChecker
hubot
xaxiclouddev.com
I don't know howto and don't have more free day.
If you are interested in filing a request for access to the CNCF Community Cluster, please fill out the details below.
If you are just filing an issue, ignore/delete those fields and file your issue.
All projects, current and interested, all members - companies and end users.
yes. Today at ciscoshipped.com. Please to update and modify to be vendor agnostic and fully open sourced.
Yes
Yes
No
Mantl, Shipped, Kubernetes, Consol, Terreform, Traefik, Cloud Foundry
400 Nodes
80%compute
2 weeks
RHEL or Centos
provide a ecsosystem for developing, creating, and collaborating cloud natives architecutures
Stuart P. Bentley
I don't really have one - I pretty much spin up a new GitHub organization for each project, so I guess that's my organization (so my org for this request would be Plushu), and my title / role would be Lead Maintainer?
Plushu
Plushu is a modular framework for building command-line interfaces, with an eye toward being used as the shell for a dedicated user on a dedicated server (ie. establishing a CLI for operating the server itself).
The Plusku ecosystem of Plushu modules uses Docker (with an eye toward Kubernetes, going forward) to build and run Twelve-Factor Apps, a la Heroku.
I currently have an open server where users can try this out running at sandbox.plushu.org (with the entry-point to submit a public key for access running at http://enter.sandbox.plushu.org), which is re-imaged daily via a Cron job on a remote machine. The Sandbox server itself is currently running on a Digital Ocean droplet I'm paying for out-of-pocket, and the cronjob is running on a Dreamhost server.
Anybody looking to self-host a Heroku-style build and deployment infrastructure for their app.
I would only use the tiniest tier of machine; I would probably be using at least three of them (one for the Pluahu Sandbox, one for the watchdog service, and one for experimentation), but not many more than that (and I could potentially work toward consolidating the Sandbox and the watchdog to one machine with two IPs, where one would be in use for the host environment and the other would point to the Sandbox VM).
The Plushu Sandbox will probably run on Ubuntu 16.04 (the current sandbox is running on 14.04, only because I haven't bothered to rebuild it since 2015). I would probably experiment with RancherOS and CoreOS as well.
If I go the sandbox-as-a-nested-VM route, I may use CentOS 7 as the host environment, with something like mist.io or oVirt to manage the VM.
All the work I do is open-source: some of my other big open source projects include the Open Profilogical Web Survey, the Tabalanche browser extension, and CouchDB in a Hurry.
The Plusku ecosystem orchestrates containerized microservices (apps and addons), and working on it uncovers new ways to approach these concepts within the space of existing developer paradigms (ie. following the use patterns established by the Heroku Toolbelt).
Is it possible to get an IPv6 address space? That could be useful for a possible extension to the Sandbox where, instead of having one global sandbox shared by any prospective developers, each user could be given a separate VM (within which they can provision containers), with its own IPv6 address.
For the sake of transparency it probably makes sense to lay down meeting note agenda and action items so folks that potentially miss a meeting can be kept up to date.
If you are interested in filing a request for access to the CNCF Community Cluster, please fill out the details below.
If you are just filing an issue, ignore/delete those fields and file your issue.
Doug
Davis
IBM
STSM
Scalability for Kubernetes, Swarm, Mesos on OpenStack
We want to see how prevalent platforms for containers would scale on
an OpenStack infrastructure. The platforms include Kubernetes, Swarm
and Mesos. As part of the study, we expect to identify bottlenecks to
fix and best practices for operating a large scale platform for
containers. We have been working on building these platforms for the
past 1.5 years. We have made some initial benchmark runs on a smaller
scale environment.
OpenStack provides cloud IaaS and Magnum is a service to deploy and
manage platforms for hosting containers on OpenStack. Currently it
supports Kubernetes, Swarm and Mesos. Each cluster is fully
integrated into OpenStack by leveraging services such as Keystone,
Nova, Neutron, Cinder, etc.
To support large scale container hosting, we need to understand the
scalability of these clusters from several perspective: many users
running multiple clusters, large cluster, launching many containers,
managing large number of containers, etc.
To study these scenarios, we have developed a set of Rally benchmarks
to run the scenarios and collect performance data. Rally automatically
measures many metrics such as success rate, launch time, etc.
Several OpenStack projects will benefit from the study: Magnum, Heat, Neutron
This should also help validate upstream container platforms: Kubernetes, Swarm, Mesos
We plan to coorelate performance data published by Kubernetes to establish
the baseline.
Yes, we currently have a presentation proposal on this data for the
next OpenStack summit in October 2016
No
No
N/A
We have contributed code in OpenStack to deploy Kubernetes, Swarm, Mesos.
We have done performance study on Kubernetes networking.
We currently have 6 patches in Rally to test various container scenarios.
We have made 2 presentations at the OpenStack Summit related to containers.
No
Two of the three members of the team are core contributors in OpenStack
One member has been contributing for the past 3 years.
50 nodes, followed by 500 nodes
50 nodes for 1 week to build OpenStack and run initial benchmarks
followed by 500 nodes for 1 week to run full scalalibity benchmarks.
With Ubuntu 14.04
It would be preferrable to have 2 network interface on the nodes, with separate networks.
This issue will (hopefully) be updated weekly to show the current CNCF community cluster allocations.
If you are interested in filing a request for access to the CNCF Community Cluster, please fill out the details below.
If you are just filing an issue, ignore/delete those fields and file your issue.
Thomas
Graf
Cilium
Cilium
Cilium addresses large scale policy enforcement and addressing for containers with the help of BPF and IPv6. We would love to run continuous regression tests of development Linux kernels to avoid regressions of Linux kernel code in particular involving BPF.
Cilium provides fast in-kernel networking and security policy enforcement for containers based on eBPF programs generated on the fly. It is an experimental project aiming at enabling emerging kernel technologies such as BPF and XDP for containers.
However, we regard the testing effort to be relevant to any container related networking solution which relies on kernel functionality.
All members with an interest in container networking or with interest in BPF technology for both instrumentation and networking.
https://github.com/cilium/cilium
Yes
Containers only
We require to run a very recent Linux kernel on the bare metal system (4.8.rcX) and we require IPv6 to be allowed on the network.
No
10+ years Linux kernel development, OVS, BPF, Docker, Kubernetes, ...
We are happy with any number of nodes we can acquire.
Ideally we can run a small number of nodes consistently to allow for regression testing as Linux kernel and BPF development progresses.
We require to run a recent Linux kernel (4.8.rcX) with BPF capabilities enabled. Docker runtime.
orchestration, microservices or some combination).
THe kernel community currently does not include any regression or automated tests which cover the special scale and performance requirement needs of cloud native workloads. The aim is to close this gap.
If you are interested in filing a request for access to the CNCF Community Cluster, please fill out the details below.
Lee
Calcote
SolarWinds
Sr. Director, Technology Strategy
Container Network Performance
Selecting the proper container network provider for a project can be difficult. We have noticed performance differences depending on the application workload.
The SolarWinds container network performance meter initiates a bandwidth test between containers and compares to the performance between the corresponding hosts. We will configure a kubernetes cluster with various network plugins (WeaveNet, Calico, Flannel, etc.) to compare performance.
Yes - network bandwidth throughput between containers compared to performance between hosts. Container network driver performance - CNI and CNM plugins.
End-users will benefit from being enlighten about how their network driver choices affect connectivity and performance of their networks. Vendors will benefit from an apples-to-apples comparison from an agnostic third-party. Kubernetes itself may benefit from an understanding of the limits of its native networking scale.
Yes, it will be open-sourced and located here - https://github.com/solarwinds/containers
Yes, absolutely.
Yes, ideally testing containers from that of the CNCF Demo project
No
No
No
20 nodes will work. 60 would be ideal in order elicit fully-distributed network behavior.
Ideally, the same compute across the nodes, so as to facilitate a uniform test. If there is variance in compute, that's fine. We'll just note this. Variance in storage is not an issue.
Ideally, 2 weeks, but 1 week will suffice.
CentOS 7 (any others will work, however)
We intend to run high-volume bandwidth and latency tests between the networked hosts.
Eli
Mallon
Streamplace Inc.
CEO
Streamplace
We're working on a cloud native compositor for live video. The core of it is a WebGL video compositor. Surrounding that core are adapters for RTMP, WebRTC, and HLS/DASH.
The incubator analogy-pitch is something like "Wordpress for livestreams" or "running your own Twitch". But that's later. For now I'm focusing on getting reliable and straightforward infrastructure for broadcasting and transcoding live video on Kubernetes.
Any of the companies that process live video in a Kubernetes cluster. Looking over the list of CNCF members, that's more than a few.
Yep, everything is Apache licensed. https://github.com/streamplace/streamplace.
A few Type 1Es would be amazing — our biggest resource hog is bandwidth. Eventually we'll want to see how high we can dial up ffmpeg's quality settings on something like a Type 2, but that can happen later.
To be clear, we don't intend to host enormous public live video broadcasts on the CNCF servers. Our clusters will be transcoding the video and broadcasting them out to YouTube/Facebook Live/Twitch/Twitter/whatever. If we are hosting our own video publicly, we'll do so through a CDN.
Container Linux Stable should be fine. Or whatever. Our current servers are running both Ubuntu and CentOS, so we're easy.
Sure —
Do a Google search for any combination of the words "live video kubernetes" and you'll get no results relevant to that at all, save for @jbeda's excellent TGI Kubernetes livestream. It gets worse with more specific terms, the fifth result for "RTMP server Kubernetes" is my npm account, so there's not a lot of resources out there for this sort of thing.
Streamplace is already publishing Helm charts for this sort of thing, including an RTMP server and a TURN server for WebRTC. I'd love to get all of that stuff polished enough to get into the main Charts repo.
Each stream involves realtime video transcoding and bandwidth up to ~50Mbps. Should be fine given the specs but it's potentially more bandwidth-intensive than most tenants.
We should eventually have a clickthrough.
If you are interested in filing a request for access to the CNCF Community Cluster, please fill out the details below.
If you are just filing an issue, ignore/delete those fields and file your issue.
Ben
Kochie
Prometheus Team
Benchmarking
We would like to build a Prometheus test setup for making performance optimizations that are not possible to test in standard CI pipelines.
Prometheus community in general.
Yes
Yes, we would use and monitor containers as part of the workload test.
None expected.
No
Prometheus Team member
50-100
2 weeks.
Ubuntu 16.04
It will improve Prometheus performance and efficiency.
If you are interested in filing a request for access to the CNCF Community Cluster, please fill out the details below.
Eugene
Zilman
CNCF
Demo
Testing scale on bare metal
The goal of this project is to demonstrate each of the technologies that have been adopted by the Cloud Native Computing Foundation (CNCF) in a publicly available repository in order to facilitate their understanding through simple deployment tooling and by providing sample applications as common-ground for conversation. This project will enable replicable deployments and facilitate quantification of performance, latency, throughput, and cost between various deployment models.
With this one we'd like to check how overlay networking performs with a high node count. Specifically we'll be benchmarking Weave and observing the behaviour with the collaboration of the Weave developers.
People interested in running Weave and Countly on larger clusters.
Yes, everything is or will be available here: https://github.com/cncf/demo
Yes.
Containers used extensively.
No
Yes
Demo
N/A
100
2 weeks
With CentOS7
It will make sure the technology demonstration ("Demo") scales from your laptop all the way to a big bare metal cluster and works as expected. Any bugs discovered will be reported upstream.
No
If you are interested in filing a request for access to the CNCF CIL, please fill out the details below.
If you are just filing an issue, ignore/delete those fields and file your issue.
If you are interested in filing a request for access to the CNCF Community Cluster, please fill out the details below.
If you are just filing an issue, ignore/delete those fields and file your issue.
Camille
Fournier
ZooKeeper
ZooKeeper
Distributed coordination service
Distributed Coordination Service
This will be used to help create a reproducible process for vetting release candidates
Anyone using Apache ZooKeeper, which is a very long list
Yes, zooeeper.apache.org
Anything that becomes a reusable artifact will be published as source code, and I certainly hope we can get something done in the next 2 months
Maybe? Maybe? Maybe?
Fully volunteer group is busy
no
TOC member for CNCF
PMC member for Apache ZooKeeper
50 nodes
open-ended
with
reproducible end-to-end and stress testing of distributed systems is a good thing for all of us
JVM
If you are interested in filing a request for access to the CNCF Community Cluster, please fill out the details below.
If you are just filing an issue, ignore/delete those fields and file your issue.
Russell
Bryant
Red Hat
Senior Principal Software Engineer
OVN Data Plane Testing
Networking data plane performance is very important, and it's often difficult to know what performance characteristics to expect with different combinations of hardware, software, and protocols in use.
We aim to do some benchmarking of OVN (from the Open vSwitch community) using the Geneve protocol as compared to other solutions. This lab has Intel X710 NICs, which support both Geneve and VXLAN offload, making it an ideal place to do this work.
Yes, we will gather networking data plane performance metrics.
Anyone interested in overlay network based solutions, particularly those based on OVS, would benefit from the additional insights gained by this work.
Yes. OVN is a part of Open vSwitch and the code is hosted at http://github.com/openvswitch/ovs. We aim to test OVN as configured by OpenStack (http://github.com/openstack/networking-ovn). We are also interested in testing with Kubernetes (http://github.com/openvswitch/ovn-kubernetes).
Yes.
Testing of ovn-kubernetes would involve containers.
Any encountered lab network or other hardware issues are the main risks.
No
I have been contributing to various open source projects since 2004. In the last couple of years I have been working on OVN (Open Virtual Network), a new network virtualization system developed by the Open vSwitch community that has been integrated with multiple systems, including OpenStack and Kubernetes.
20 (or less is fine, probably 5 minimum)
any (primary requirement is the Intel X710 NIC, which appears to be in both compute and storage flavors)
2 weeks
CentOS 7, though ideally also with the option to re-provision the machines using an OpenStack deployment tool.
OVN is a new network virtualization option that works with Kubernetes. Having a better understanding of the performance of OVN and its use of Geneve will help better inform choices and further development in this area.
No
If you are interested in filing a request for access to the CNCF Community Cluster, please fill out the details below.
If you are just filing an issue, ignore/delete those fields and file your issue.
Xiang
Li
CoreOS Inc.
etcd
Distributed reliable key-value store for the most critical data of a distributed system
Distributed reliable key-value store for the most critical data of a distributed system
Improve reliability, scalability of etcd
Kubernetes and other distributed system in general.
Yes. https://github.com/coreos/etcd
Yes.
Yes.
No.
No
Maintainer of etcd, Kubernetes.
80
open ended
etcd is the critical part of distributed system, including orchestration and microservices.
Making etcd more reliable and scalable improves other applications that rely on it.
If you are interested in filing a request for access to the CNCF CIL, please fill out the details below.
If you are just filing an issue, ignore/delete those fields and file your issue.
Rom
Bob
CICD XYZ
DevOps
Infrastructure management
Infrastructure management toolset for opensource projects CICD solutions not requiring hosted CI services for SDLM
Yes, TBD
Everyone,
Due to the expansion process going on we will be scheduling a maintenance window soon. We are not yet sure about the exact schedule but it will probably take time around end of January / beginning of February and last up to a few days.
Because of this, no new requests will be accepted until after the maintenance window. If required, the current requests will be allowed to extend their work week by week until we are sure about the date.
Feel free to reach us here in case of questions!
Hello,
I am just curious. What is the CNCF doing to make sure there is tangible benefit(s) from use of the cluster? I see an approval process, but not really a follow up in what was improved about the software involved. Not trying to call any project out, just trying to close the loop so others can reference this as good use and an overall benefit.
I think maybe just linking back to an issue in the various projects would be a good start. (ex. software X has never ran on 500+ machines, it be good to see it work). I also realize it might not lead to some change, but should at least show a decision was made based on its usage. I can appreciate how much time(aka money) and effort goes into running clusters.
If I missed something, I apologize.
Ron
If you are interested in filing a request for access to the CNCF CIL, please fill out the details below.
If you are just filing an issue, ignore/delete those fields and file your issue.
Kubernetes in hybrid environments, please see https://kccncna17.sched.com/event/CU7L/kuberneters-in-hybrid-environments-using-cloud-interconnect-a-marc-chisinevski-f5-networks
Istio/Envoy/Kubernetes
https://github.com/F5Networks/k8s-bigip-ctlr/
Type 1E
Public Internet
https://sourceforge.net/projects/assetmng/files/assetmng/
https://sourceforge.net/projects/assetmng/files/assetmng/v1.0/Asset_Management_and_Risk_Assessment___Functional_and_Technical_Specifications___plus_Linux_installation.pdf/download
Ease migration to managed Kubernetes environements (Google Container Engine, Azure Container Services etc) and enabling hybrid deployments.
If you are interested in filing a request for access to the CNCF Community Cluster, please fill out the details below.
Latitia
Haskins
IBM/ITPC/BlockChain
software engineer
Getting 100 fabric-peer nodes running running in an hyperledger fabric network
We want to verify that we can run a high number of peers in many (greater than 10) organizations. Ideally, we would be able to test this with each peer being in its own organization. We also want to reconfigure this to have more than 1 peer in each organization with 1 peer having access to a channel and the other not having access as well as the number of peers allowed in a single organization and possibly more.
This test will allow us to find possible holes when there is a high number of organizations and peers with a certain number of channels, with only certain peers having access to a specific chaincode.
yes, We would like to be able to track
Yes, https://github.com/hyperledger/fabric
Yes, for now, we can put the results in a README
Yes, we have docker images.
Not that I'm aware of at this time.
No.
This is for the Hyperledger-Fabric open source project.
125 nodes.
compute nodes, but we will want to look at both TLS and non-TLS setups, so we will need to be able to save certificates for potentially 100 different nodes on each node. The specified 400GB should be more than enough.
2 weeks
With - Ubuntu 16.04
This is testing scalability capabilities for the fabric network. It will give us more concrete indications of where the network bottlenecks are located and the extent of those bottlenecks. Using this setup will not limit us to the possibility of our local system limitations, but fabric network limitations.
As stated previously, we use docker images for the different containers. We would like to be able to view the containers' logs and log into certain nodes, if necessary.
Ihor
Dvoretskyi
CNCF
Developer Advocate
All existing (and related) CNCF projects - https://www.cncf.io/projects/
Everyone would!
Yes. GitHub links are available here - https://www.cncf.io/projects/
1 * TYPE 1
Ubuntu 16.04 LTS
https://github.com/idvoretskyi
https://github.com/kubernetes/community/blob/master/sig-product-management/README.md
https://github.com/kubernetes/community/blob/master/sig-openstack/README.md
https://www.cncf.io/blog/2017/09/18/meet-cncfs-newest-developer-advocate/
Not yet.
Documentation on cluster usage should probably live here, I've been looking for it but have been unable to find it.
/cc @rrati @jeremyeder
If you are interested in filing a request for access to the CNCF Community Cluster, please fill out the details below.
If you are just filing an issue, ignore/delete those fields and file your issue.
Omer
Katz
N/A
N/A
Adding NUMA-aware allocator to RethinkDB's buffer cache
Databases have high memory throughput even in clusters of modest sizes. Data is usually cached in memory and the faster it is read the faster the database is.
On NUMA-enabled hardware, databases can swap in a way that greatly degrades performance. See https://blog.jcole.us/2010/09/28/mysql-swap-insanity-and-the-numa-architecture/
The project will use the memkind to introduce NUMA aware allocations to the buffer cache.
Yes, I'm going to use the YCSB benchmark provided here. This will ensure everything is working normally and that performance is enhanced.
The RethinkDB project and its users
Yes. https://github.com/rethinkdb/rethinkdb
Yes absolutely. A PR will be issued to the RethinkDB repository.
I do not need containers. I may test RethinkDB on containers as well to see the effect of my changes on the performance of containerized RethinkDB.
Not that I know of.
No
I maintain Celery a task manager written in Python, MongoEngine which maps Mongo documents to Python objects and oauthlib which is the standard OAuth1/2 implementation for Python.
20
compute
2 weeks if possible. I'd like to repeatedly refine the performance and ensure we're getting the most out of the hardware.
Ubuntu.
I need both machines that are NUMA-enabled and machines that are not.
I'm not entirely certain if this is the appropriate place to put this, but I believe it will leverage these resources.
The CNCF hosts a fair number of different projects which are controlled by a number of different vendors. The purpose of this request is to start the conversation about the establishment of a shared infrastructure (vendor neutral) area that can be used as a test bed for these projects. This ensures that releases have to go through a common set of shared resources that no-one vendor can explicitly control but they can all contribute to. ~= apache-infra
/cc @caniszczyk @spiffxp
If you are interested in filing a request for access to the CNCF CIL, please fill out the details below.
If you are just filing an issue, ignore/delete those fields and file your issue.
Rob
Hirschfeld
RackN
CEO
Digital Rebar
Automates physical network provisioning
RackN is a CNCF member. The service works on Packet.net
We're not sure who else is consuming software
Yes. http://Rebar.digital
Code is github.com/digitalrebar/provision
We're talking about PROVISION only, not the v2 container stuff.
Type 0s. 10 - 15 machines per build for about 15 minutes per commit.
CustomIPXE exclusively
For CNCF, we've been working on Kubespray and Kubernetes
We have significant activity with OpenStack in the past
We have been involved in a number of projects in monior ways including Consul, Terraform, BOSH, Golang and Chef.
Digital Rebar is the cloud native physical provisioning alternative. This testing ensures that the community profiles work consistently. That helps users who want to use Kubernetes and other CNCF projects on bare metal deployments.
We're already using Packet for this open source testing, so we'd be able to give you a very quick and visible win on using the infrastructure to support an open project with CNCF ties.
If you are interested in filing a request for access to the CNCF CIL, please fill out the details below.
If you are just filing an issue, ignore/delete those fields and file your issue.
VIJAY
PRAKASH
CGE
SC
SC
Migration project for application from on prem to cloud
Small and Mid business members
I haven't decided one, but would start with something small
Linux and Windows
Vm to Containers
If you are interested in filing a request for access to the CNCF Community Cluster, please fill out the details below.
If you are just filing an issue, ignore/delete those fields and file your issue.
Frank
Scholten
Container Solutions
Senior Software Engineer
Mesos Elasticsearch
Deploying and running Elasticsearch on a Mesos cluster
Deploying and running Elasticsearch on a Mesos cluster
Anyone using Mesos and/or Elasticsearch
Yes. https://github.com/mesos/elasticsearch
Yes
Yes. Mesos Elasticsearch can run in Jar mode or in Docker mode by specifying a flag when running the scheduler.
Contributors are busy
No
Committer on Apache Whirr (now in attic). Committer on Apache Mahout (currently inactive). Creator of minimesos: https://github.com/ContainerSolutions/minimesos
100
One week
With a Linux operating system that supports Docker
End-to-end testing of deployment, scaling and failover scenarios will help us improve the quality of Mesos Elasticsearch framework and provide information that is useful for other projects.
We will be running on Mesos
If you are interested in filing a request for access to the CNCF Community Cluster, please fill out the details below.
If you are just filing an issue, ignore/delete those fields and file your issue.
Andres
Garcia Garcia
Fujitsu EST
Senior Software Engineer
CNCF Service Broker API
Definition of an standard Service Broker API
Task force to define the Service Broker API in the context of the CNCF activities
No
Service providers interested on implementing a standard API for service provisioning.
Yes
No
No
No
Currently on the process to contribute to https://github.com/servicecatalog/development, currently developing proof of concept prototypes for the CNCF Service Broker API workgroup.
20
3 months
With
We want a Kubernetes cluster to test the different PoC prototypes that we will develo for the Service Broker workgroup in the following months.
If you are interested in filing a request for access to the CNCF Community Cluster, please fill out the details below.
If you are just filing an issue, ignore/delete those fields and file your issue.
Developer automation for fundamental, "can be anywhere" infrastructure is a critical component of cloud native computing. To date, most cloud native computing relies heavily on primitives of multi-tenant public clouds or virtualization based private clouds. Packet is eager to refine its software stack for use in on premise environments to power cloud native applications regardless of location. At Packet, we operate a public cloud that is used by many CNCF members to test, build or operate cloud native stacks without virtualization or overlay networking as a requirement. We envision that much of this could happen on the CNCF cluster.
We would like to install and operate a Packet private rack in the CNCF cluster. This would include both server and switch automation.
We are interested in achieving baseline capabilities, including provisioning, deprovisioning, healthchecking and metering of physical compute instances. We would use our own Docker-based canary testing for simulating thousands of provisions and measure the success rates.
Nearly all members could leverage this. However, members known to be comfortable with the Packet API and meta data service include CoreOS/Quay, Mesosphere DC/OS, Rancher, Apprenda, ContainerSolutions, Deis, Diamanti, Nginx, Portworx, StackPoint, Mirantis, Tigera, and Virtuozzo.
No, it is not currently open source. Our goal would be to evaluate the usefulness of open sourcing any of our core automation.
We agree in publishing in our results.
Yes, our stack currently runs 100% in Docker (on Rancher).
Yes, the network architecture and access to control plane, IP space, routing level details, etc are unclear and could affect if this would be successful or not.
No.
We are large supporters of the Cloud Native and Open Source communities in general, regularly providing 100-1000 bare metal compute nodes to cloud native projects for load and scale testing, build operations and compatibility work.
100 nodes, or whatever is in a logical switch group.
2 weeks would probably be insufficient for our use given past experiences with other infrastructure environments. Our goal would be to a) complete Packet private rack ingestions and b) offer use of the underlying compute resources to CNCF projects via the Packet API and metadata service (using a special CNCF facility code).
No operating system. We would need full BMC access, bios level control (particularly if we were to test trusted computing), and privileged switch control using netconf style access.
Our goal is to showcase that cloud native computing can leverage fundamental bare metal compute (literally nothing else), be performed in any location and offer the same convenient DevOps interfaces that users leverage today (e.g. terraform, ansible, native cloud-init, elastic addressing, etc).
If you are interested in filing a request for access to the CNCF Community Cluster, please fill out the details below.
If you are just filing an issue, ignore/delete those fields and file your issue.
Howard
Huang
Huawei
Senior Open Source/Standard Community Manager
OpenSDS Kubernetes Integration Testing
Storage is an important part of functionality for Kubernetes, and currently there are multiple in-tree and out-of-tree volume plugins developed for Kubernetes, but it is hard to know whether the integration of storage provider is functioning as it should be without testing it. Moreover for storage that is provided from cloud provider (e.g OpenStack), there should be functional testing to prove the integration's actually working.
OpenSDS is a fellow Linux Foundation collaborative project and aims to provide a unified storage service experience for Kubernetes. OpenSDS could provide various types of storage resource for Kubernetes (e.g OpenStack block and file service, baremetal storage device), and we want to use CNCF Cluster to :
Yes we need to measure the storage performance (e.g IOPS, latency) to see if the integration is functioning properly.
Kubernetes's storage-sig will get a better view on the future storage evolution. End-users could also have first-hand knowledge of the Kubernetes Storage prowess
Yes, please refer to the source code at https://github.com/opensds/opensds
Yes, and we will also probably need to feedback the result to the OpenStack community as well.
Yes we are testing with container clusters managed by Kubernetes.
The storage interface is not well defined in Kubernetes at the moment, and any changes in the upstream will likely break the testing. We will try to work with storage sig as much as possible
No
I've been working in the OpenStack and OPNFV community for about 3 years. My stats could be found at stackalytics website
20 or less (maybe 4 or 5 node is good enough)
compute and storage node
2 weeks
Ubuntu
This testing could help with orchestration especially in the storage area.
No
Hi, it would be useful if cluster users could have (read-only) visibility into:
Without this, debugging performance, link issues, driver/firmware issues at the host level becomes very difficult. Cluster users could make more progress if they had such visibility and would reduce turn-around time for resolving issues.
A declarative, efficient, and flexible JavaScript library for building user interfaces.
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google ❤️ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.