GitLab

Signed-off-by: Spyros Trigazis <spyridon.trigazis@cern.ch>

* add ip_family_policy which defaults to dual_stack
* expose portal_network6_cidr,pods_network6_cidr as a labels
  keeping the same default cidr
* drop EndpointSlice, IPv6DualStack which are True anyway since
  1.18 and 1.21 respectively

closes https://gitlab.cern.ch/kubernetes/project/-/issues/300

Signed-off-by: Spyros Trigazis <spyridon.trigazis@cern.ch>

master_availability_zones is a comma_delimited_list of
availability_zones. master-0 gets az master_availability_zones[0],
master-N gets az master_availability_zones[N].

If the labels availability_zone is set all nodes (master and worker)
will use this availability zone, for backwards compatibility.

Change-Id: I5c7c0b0819fe350dce58ab8175466eda9e285c8e
Signed-off-by: Spyros Trigazis <spyridon.trigazis@cern.ch>

ignore-volume-az is an option specific to cinder-csi driver.
It needs to be set if compute and volume AZs are distinct.

OS-15285

Install monitoring by using the updated kube-prometheus-stack
helm chart.
Signed-off-by: Diogo Guerra <diogo.filipe.tomas.guerra@cern.ch>
Change-Id: I12ccfed3308d02600ccf09c5a9aa7e92a2eb7348

Calico does not support CALICO_IPV6POOL_IPIP,
we need to be able to define the IP encapsulation
to make sure connectivity works across subnets
(subnets of the VMs).

We can define the default IPPOOls by hand.

Change-Id: I947f06b5f96614a10a8e9f52e9728840cf323d7e
Signed-off-by: Spyros Trigazis <spyridon.trigazis@cern.ch>

Move out of auth resources v1beta1, add csidriver definition to cvmfs
script since we moved cephfs csi to helm.

Restart network manager until we get an ipv6.

Spyridon Trigazis authored Nov 16, 2021 and Ricardo Rocha committed Nov 23, 2021

Change-Id: Idcb57107a3c3c6c1a3fb98286f7b211017aca968
Signed-off-by: Spyros Trigazis <spyridon.trigazis@cern.ch>

Install traefik v2 from cern metachart greater than 0.9.x

This commit is an adaptation of the upstream commit
https://review.opendev.org/c/openstack/magnum/+/697071

Signed-off-by: Diogo Guerra <diogo.filipe.tomas.guerra@cern.ch>
Change-Id: I4b5b3238f405c65b85b916e4fbe71cb37d5b548e

Diogo Filipe Tomas Guerra authored Oct 28, 2021 and Ricardo Rocha committed Nov 15, 2021

Edit the logic of magnum to install nginx
by using the cern metachart.
Signed-off-by: Diogo Guerra <diogo.filipe.tomas.guerra@cern.ch>
Change-Id: I0f5a4fbce5ea031abca5ee4bdc53a0c31fa06274

This commit adds following labels:
* snapshot_controller_enabled
* snapshot_controller_version
* snapshot_validation_webhook_enabled
* snapshot_validation_webhook_version

OS-13997

OS-14396

We run kube*, etcd and heat-container-agent with
podman. We need to move from libpod.conf to
containers.conf

Keep the limit to 50M.

OS-14733



Change-Id: Id0a6a2b453f2ad28696783a41c184d6d526d8a66
Signed-off-by: Spyros Trigazis <spyridon.trigazis@cern.ch>

Spyridon Trigazis authored Jul 12, 2021 and Ricardo Rocha committed Jul 15, 2021

cern_chart_values is a bsae64 --wrap=0 encoded string
that stores the helm values of the cern metachart.
https://gitlab.cern.ch/helm/releases/cern-magnum/-/blob/f60ddc94c7264d3968b07f62048a17d8cbbfdeea/values.yaml



If the values exist labels like monitoring_enabled, logging_producer
will have no effect since they should be defined in the values file.

The values will be decoded, indented and writen in the ConfigMap of the
install-cern-chart job.

If cern_chart_values is "" (the default value), the cern_chart will be
installed with the values already in helm/cern-chart.sh.

The values can contain bash variables in yaml which will be replaced
for backwwards compatbility.

eg in:
nvidia-gpu:
  enabled: ${NVIDIA_GPU_ENABLED}

${NVIDIA_GPU_ENABLED} will be replaced by the value of the
nvidia_gpu_enabled label.

Change-Id: I93319d2a42fc52fe3d9966e04008eac34eb7e927
Signed-off-by: Spyros Trigazis <spyridon.trigazis@cern.ch>

Thomas George Hartland authored Jun 07, 2021 and Ricardo Rocha committed Jul 15, 2021

The public discovery.etcd.io service very rarely
returns non 200 responses due to unknown errors.

From what I've seen the error is always HTTP 400
and the message is "unable to generate token".

Retrying the request should prevent this error
from blocking the cluster creation.

Feilong Wang authored Apr 08, 2020 and Ricardo Rocha committed Jul 15, 2021

upstream: https://review.opendev.org/c/openstack/magnum/+/718296

In commit I1a75f1bf12747508a3497293650d3cc668202de6 the worker node
is missed to add the docker storage support. And the current systemd
unit is not really working. So this patch fixes it by removing the
hardcode for /dev/vdb and using xfs instead of ext4 (the same way
for Fedora Atomic) to make it simpler and solid.

Task: 39331
Story: 2005201

Change-Id: I4c465664eb19f1992df95750dd7b2d99688c6cae
(cherry picked from commit c2439ca1)

Spyridon Trigazis authored Jul 12, 2021 and Ricardo Rocha committed Jul 15, 2021

1.22 drops apiextensions.k8s.io/v1beta1 and moves to apiextensions.k8s.io/v1
https://github.com/kubernetes/kubernetes/pull/102197

CRDs generated from https://github.com/projectcalico/calico/tree/v3.15.0/


make values.yaml
cd _includes/charts/calico/
helm template .

Change-Id: I8131d59b9983fb88bcbed7dfe4b9ff938f2a7545
Signed-off-by: Spyros Trigazis <spyridon.trigazis@cern.ch>

On FCOS 34 podman is injecting an extra line into
/etc/hosts in the kubelet container that the kubelet
then passes into hostNetwork pods, which is causing issues.

https://its.cern.ch/jira/browse/OS-14303

FCOS 34 enables cgroupsv2 by default and comes with an
updated ingition that supports kernelArguments.
https://github.com/coreos/fedora-coreos-docs/blob/main/modules/ROOT/pages/kernel-args.adoc

Added in igntion 2.10.x
https://github.com/coreos/ignition/releases/tag/v2.10.1



Add a label to set the igntion payload version.
* 3.3.0-experimental for FCOS34
* 3.0.0 for FCOS31
  * We will see a warning with ignition ignoring the kernelArguments
    section.

Change-Id: I49257389ca19bf40de9231cc98e3503b8c9bd7ea
Signed-off-by: Spyros Trigazis <spyridon.trigazis@cern.ch>

Ricardo Rocha authored Jun 22, 2021 and Ricardo Rocha committed Jun 22, 2021

Required for Fedora CoreOS 34 which defaults to v2.

https://github.com/kubernetes/kubernetes/pull/6380
https://github.com/kubernetes/kubernetes/pull/91630



Change-Id: I6ccd173841edce5e1a0605b6abaf6fe8d14d619f
Signed-off-by: Spyros Trigazis <spyridon.trigazis@cern.ch>

This was superseded by ceph-csi-cephfs attribute:
https://gitlab.cern.ch/cloud-infrastructure/magnum/-/merge_requests/168/diffs

Diogo Filipe Tomas Guerra authored Aug 25, 2020 and Robert Vasek committed May 11, 2021

When CERN_METACHART_ENABLED is true cephfs-csi is installed from
umbrella chart instead of the magnum existent scripts.

If this option is false and the user has set CEPHFS_CSI_ENABLED
the components are installed with the magnum scripts guaranteeing
that no backwards compatibility is lost.

Change-Id: I6d3312a43ad0470fae9ca1fd84173b14d0bf2a77

Robert Vasek authored May 11, 2021 and Ricardo Rocha committed May 11, 2021

Added labels manila_csi_enabled and manila_csi_version.

OS-13794

OS-13799

Diogo Guerra authored Mar 26, 2021 and Diogo Filipe Tomas Guerra committed Apr 27, 2021

Upstream: https://review.opendev.org/c/openstack/magnum/+/783390



* Traefik version updated from v1.7.19 to v1.7.28
* Force secure connections to use TLSv1.2 or greater

Change-Id: I65561358113952e3f60dc488b35ee8fa8f8da740
Signed-off-by: Diogo Guerra <diogo.filipe.tomas.guerra@cern.ch>