Topics

RKE HA collaboration for Kubernetes/Helm/Docker alignment #integration #oom #documentation


Michael O'Brien <frank.obrien@...>
 

Gary,

   Hi, we (OOM) would like to continue our collaboration with yourself, Brian, Yang, Alain, Mike, Sylvain, Morgan, Eric, …and others inside devops teams on getting an HA deployment of RKE aligned via a script/template/documentation-procedure up for RC0.

   The goal/deliverables will be as we did together before for Rancher 1.6.25 in Casablanca – but this time for Dublin around RKE as a replacement for Rancher.

  • Community based RI for RKE to be used for staging/testing and production readiness
  • Alignment on the k8s/helm/docker versions

 

   We have the following merged collocated/single-node script based on the 2 month old RKE 0.1.17 that we will be upgrading to RKE 2.1.x to leverage fixes in the move from Docker 17.03/18.06 to 18.09, Helm 2.9.1 to 2.12.5 and Kubernetes 1.11.5 to 1.13.5 – some of these moves solve critical vulnerabilities.

    https://git.onap.org/oom/tree/kubernetes/contrib/tools/rke/rke_setup.sh#n98

 

    Tracking via https://jira.onap.org/browse/OOM-1598

    Preliminary docs to be rewritten for RTD via https://wiki.onap.org/display/DW/OOM+RKE+Kubernetes+Deployment

 

   I understand you are actively testing a larger cluster and script and distributing the 3 roles (controlplane, worker, etcd) – which I currently run on a single node.   We would like to align with your proposed VM distribution of these 3 roles (like we did for rancher on the 13+1 cluster).  We will then do a 2 step distribution of the yaml to documentation for RC0 with full script retrofit for RC1+

   Did a quick check in git and I will look through your https://git.onap.org/integration/tree/deployment/heat/onap-rke for the yaml/role profile.

 

   Thank you

   /michael

This email and the information contained herein is proprietary and confidential and subject to the Amdocs Email Terms of Service, which you may review at https://www.amdocs.com/about/email-terms-of-service


Gary Wu
 

Hi Michael,

 

I have updated the integration scripts to use the following versions (also updated OOM-1598 with this info):

 

·       ubuntu 18.04

·       docker 18.09.5

·       rke 0.2.1

·       rke kubernetes_version: v1.13.5-rancher1-2

·       helm 2.12.3

·       kubectl 1.13.5

 

helm 2.12.5 did not appear to exist, so I lowered it to the latest 2.12.x that I could find which was helm 2.12.3.

 

Our k8s cluster layout is currently as follows:

·       12 x (16 GB RAM, 8 vCPUs) running the worker role

·       3 x (8 GB RAM, 4 vCPUs), each running both controlplane and etcd roles

·       1 x (8 GB RAM, 4 vCPUs) functioning purely as the shared NFS server for /dockerdata-nfs/.

 

Thanks,

Gary

 

From: Michael O'Brien [mailto:Frank.Obrien@...]
Sent: Tuesday, April 23, 2019 11:11 AM
To: onap-discuss@...; Gary Wu <gary.i.wu@...>; Yang Xu (Yang, Fixed Network) <Yang.Xu3@...>; Brian Freeman (Confluence) <noreply@...>; Mike Elliott <Mike.Elliott@...>
Cc: Soleil, Alain <Alain.Soleil1@...>; sylvain.desbureaux@...; sofia.wallin@...; James MacNider <James.MacNider@...>; Prudence Au <Prudence.Au@...>; Tcherniavski, Igor <igor.tcherniavski@...>; michel.nguyen@...; eric.debeau@...; morgan.richomme@...; Adolfo Perez-Duran <adolfo.perez-duran@...>; Beejal Shah <Beejal.Shah@...>; m.ptacek@...; p.ospaly@...
Subject: RKE HA collaboration for Kubernetes/Helm/Docker alignment #oom #integration #documentation

 

Gary,

   Hi, we (OOM) would like to continue our collaboration with yourself, Brian, Yang, Alain, Mike, Sylvain, Morgan, Eric, …and others inside devops teams on getting an HA deployment of RKE aligned via a script/template/documentation-procedure up for RC0.

   The goal/deliverables will be as we did together before for Rancher 1.6.25 in Casablanca – but this time for Dublin around RKE as a replacement for Rancher.

-          Community based RI for RKE to be used for staging/testing and production readiness

-          Alignment on the k8s/helm/docker versions

 

   We have the following merged collocated/single-node script based on the 2 month old RKE 0.1.17 that we will be upgrading to RKE 2.1.x to leverage fixes in the move from Docker 17.03/18.06 to 18.09, Helm 2.9.1 to 2.12.5 and Kubernetes 1.11.5 to 1.13.5 – some of these moves solve critical vulnerabilities.

    https://git.onap.org/oom/tree/kubernetes/contrib/tools/rke/rke_setup.sh#n98

 

    Tracking via https://jira.onap.org/browse/OOM-1598

    Preliminary docs to be rewritten for RTD via https://wiki.onap.org/display/DW/OOM+RKE+Kubernetes+Deployment

 

   I understand you are actively testing a larger cluster and script and distributing the 3 roles (controlplane, worker, etcd) – which I currently run on a single node.   We would like to align with your proposed VM distribution of these 3 roles (like we did for rancher on the 13+1 cluster).  We will then do a 2 step distribution of the yaml to documentation for RC0 with full script retrofit for RC1+

   Did a quick check in git and I will look through your https://git.onap.org/integration/tree/deployment/heat/onap-rke for the yaml/role profile.

 

   Thank you

   /michael

 

This email and the information contained herein is proprietary and confidential and subject to the Amdocs Email Terms of Service, which you may review at https://www.amdocs.com/about/email-terms-of-service