v0.15.0

v0.15.0

This release include a much much better documentation and some nasty
bugs has being fixed, it does not implement new features, except for
externa library integration for webhooks and the pprof integration
into the project.

NOTE: This is the last minor release for NetworkManager 1.20, next minor release
will use NetworkManager >= 1.22

Features:

• Adding profiler: (#386)
• Integrate kube-admission-webhook module (#381)

Bugs:

• Use policy generation to calculate overall condition. (#407)
• don't retry the configuration (#388)
• wait for scc resource (#393)
• Default to 0 if numberAvailable is not present (#380)

Docs:

• add supported networkmanager version to README (#403)
• Add a link to the vlan and ip configuration user guide (#382)
• add supported networkmanager version to README (#403)
• docs: 103 troubleshooting (#395)
• docs: 102 Configuration (#392)
• doc: add ubuntu netplan configuration (#400)
• don't retry the configuration (#388)

docker pull quay.io/nmstate/kubernetes-nmstate-handler:v0.15.0

v0.14.0

v0.14.0

This release include a mutating admission webhook that will ensure that
NodeNetworkConfigurationPolicy conditions are reset after being create/modified
by clients, that prenvets a race condition between kubectl apply and kubectl wait

Another interesating feature is a new column at kubectl get nncp/nnce that shows the status:

$ ./kubevirtci/cluster-up/kubectl.sh get nncp
NAME                                     STATUS
example-nodenetworkconfigurationpolicy   SuccessfullyConfigured

The project now also use a vendored golang version that get installed if is not already
present ander ./build. Also the features are tested with k8s 1.17.

Features:

• Run webhook only on master nodes (#361)
• expose Reason on NNCP and NNCE list (#374)
• Use k8s 1.17.0 provider (#339)
• Pin and install go from go.mod version (#352)
• Add conditions reset mutating webhook (#324)
• Use enactments matching state for policy conditions calculation (#306)

Bugs:

• Count seconds instead of nano for default gw retrieval (#366)
• Use enactments matching state for policy conditions calculation (#306)
• Call generators before handler build (#344)
• drop user guides root (#363)
• drop outdated guides and merge developer guide (#358)

Docs:

• doc: 101 state (#368)
• Add user guide for vlan and dynamic ip policy (#354)
• make readme straight to the point (#367)
• Add CI infrastructure to contributing page. (#364)
• Renamed docs/developer-guide.md to docs/CONTRIBUTING.md (#353)

docker pull quay.io/nmstate/kubernetes-nmstate-handler:v0.14.0

v0.13.0

v0.13.0

This release add a policy overall conditions that represent
the cluster state of a policy, also operator-sdk has being bump
to 0.12.0

Features:

• Add overall policy conditions (#300)
• Add desired state field to enactments (#287)
• Add Matching condition to enactments (#285)
• Move enactments to a proper CRD (#279)
• Add LastSuccessfulUpdate field to NodeNetworkState (#280)
• Add Progressing state to enactment (#271)
• Decouple configuration and reporting (#262)
• Use go modules and operator sdk 0.12.0 (#246)

Bugs:

• Retry default gw retrieval on error (#328)
• Re-read policy at retryOnConflict (#319)
• Force Get enacment after update (#307)
• Fix monitoring logs (#277)
• Override nmstate State openapi type (#268)
• Upgrade CRDs to operator-sdk 0.12.0 (#266)
• Remove logs from metrics after operator-sdk upgrade (#263)

Docs:

• Use dhcp for cluster-wide policies (#308)

docker pull quay.io/nmstate/kubernetes-nmstate-handler:v0.13.0

v0.12.0

• Workaround for nmstate mainloop timeout which is too short.
• Use nsmtate RPM released under 0.1.1 version. This should reflect the current
  master.

docker pull quay.io/nmstate/kubernetes-nmstate-handler:v0.12.0

v0.11.0

Use latest nmstate obtained from master

The version 0.8.0 we used until now is pretty old and contains some major bugs.
We are moving to use the current master until nmstate publishes a new stable
release.

docker pull quay.io/nmstate/kubernetes-nmstate-handler:v0.11.0

v0.10.0

Increase resilience and force vlan_filtering at bridge

The NetworkManager < 1.20 present some race conditions that end with
nmstatectl set with transitory failures, they can be overcome by some
retry + sleep mechanism implemented in this release.

A rollback mechanism has being implemented to go back to previous network
configuration in case of error at running the vlan-filtering script or pinging
default gw.

The vlan-filtering script was not setting the pvid untagged flag at the bridge
itself.

docker pull quay.io/nmstate/kubernetes-nmstate-handler:v0.10.0

v0.8.0

Increase nmstate set timeout to 60 seconds

The nmstate main loop timeout is harcoded to 20 seconds this is problematic
at some scenarios, this release increment it to 60 seconds.

docker pull quay.io/nmstate/kubernetes-nmstate-handler:v0.8.0

v0.7.0

Support untagged traffic at linux bridges

There were some issues sending untagged traffic out of the bridge with
#163 we fix that.

Also:

• Add policies to documentation
• Use range 2-4094 for vlan filtering at linux bridges

docker pull quay.io/nmstate/kubernetes-nmstate-handler:v0.7.0

v0.6.0

Upgrade nmstate to 0.0.8 version

There is a problem deleting bridges at nmstate nmstate/nmstate#445 this upgrade nmstate
so we fix that

docker pull quay.io/nmstate/kubernetes-nmstate-handler:v0.6.0

v0.5.1

Update NodeNeworkState desiredState only on changes

Before this release it was updateing desiredState at the same time that we refresh
currentState leaving to connectivity issues. Now currentState is update at different
controller to refresh it so desiredState is only applyed if it's really changed.

docker pull quay.io/nmstate/kubernetes-nmstate-handler:v0.5.1