Calico: NodeNetworkUnavailable condition fix not included in v3.4.0 as release notes suggest

[ffilippopoulos]

According to release notes for v3.4.0 the following fix should be included:

For compatibility with the GCE cloud provider, Calico now clears the Kubernetes NodeNetworkUnavailable condition. node #89 (@aarnaud)

should be included.
The commit for the above issue seems to only be included in release-v3.3 branch.
https://github.com/projectcalico/node/pull/89/commits/d2f4efca880558c11bb542e84077bc9ac705fcfa

Expected Behavior

Deploying calico-node 3.4.0 to kube-gcp cluster should mark NodeNetworkUnavailable as false on nodes that calico-node is running.

Current Behavior

Had to downgrade to quay.io/calico/node:v3.3.2 image to succeed the above

Possible Solution

Either update the release notes or add the fix in 3.4

Steps to Reproduce (for bugs)

Deploy calico v3.4.0 on a gcp kubernetes cluster.

Context

Holding us from updating to calico v3.4 and use its features like bgp advertisement of services address range

Your Environment

Kubernetes cluster running in gcp.

[caseydavenport]

@ffilippopoulos thanks for raising - yes, this was somehow marked for v3.4 but instead added to v3.3..

We will release a Calico v3.4.1 which addresses this. I've adjusted the Calico v3.4.0 release notes to align with reality.

[george-angel]

@caseydavenport I think we can skip v3.4.0 and go straight to v3.5.0 (released on the 26th), can you confirm v3.5.0 has this fix? Thank you.

[caseydavenport]

v3.5.0 does have the fix, yep!

[ffilippopoulos]

closing since both release notes are updated and v3.5.0 includes the commit