k3s/cmd
Kubernetes Submit Queue da932c19c8
Merge pull request #58769 from pmichali/node-cidr-fix
Automatic merge from submit-queue (batch tested with PRs 59394, 58769, 59423, 59363, 59245). If you want to cherry-pick this change to another branch, please follow the instructions <a href="https://github.com/kubernetes/community/blob/master/contributors/devel/cherry-picks.md">here</a>.

IPv6: Ensure calculated node CIDR size for pod subnets is valid

With IPv4, the node CIDR prefix is set to /24, which gives 256 pods per node
and 256 nodes, when assuming a /16 is used for the pod subnet.

For IPv6, the node CIDR prefix, is hard coded to /64. This does not work,
because the pod subnet prefix must be /66 or higher and must be a larger subnet
(lower value) than the node CIDR prefix.

In addition, the bit mask used to track the subnets (implying the number of
nodes), can only handle 32K entries, so the difference between pod subnet
prefix and node CIDR prefix cannot be more than 16 (bits).

To address this, the following algorithm is proposed to provide as many pods
per node as possible, and not exceed the number of nodes.

If the pod subnet prefix is from /66 to /104, we'll set the node CIDR prefix
to 16 more, so that the bit map is not exceeded. The rest of the bits will be
for pods per node.

If the subnet prefix is from /105 to /112, we'll split the available bits
between what is used for the nodes and what is used for the pods per node.
This will give a node CIDR prefix from /116 to /120.

If the subnet prefix is from /113 to /119, we'll do like IPv4 and ensure that
there are 256 pods per node, and the remaining bits will be for the nodes,
giving a node CIDR prefix of /120 always. This supports a limited number of
nodes, in some cases.

If the subnet prefix is /120 to /128, we don't have enough bits and will set
the node CIDR prefix to be the same as the pod subnet prefix. This will cause
a falure later, when it tests that the pod subnet prefix is larger than the
node CIDR prefi.



**What this PR does / why we need it**:

**Which issue(s) this PR fixes** *(optional, in `fixes #<issue number>(, fixes #<issue_number>, ...)` format, will close the issue(s) when PR gets merged)*:
Fixes #58766 

**Special notes for your reviewer**:

**Release note**:

```release-note
NONE
```
2018-02-06 21:34:40 -08:00
..
clicheck Autogenerate BUILD files 2017-12-23 13:12:11 -08:00
cloud-controller-manager uniquify resource lock identities 2018-01-18 11:31:43 -05:00
controller-manager/app/options Seperate loop and plugin control 2017-12-18 10:58:45 -08:00
gendocs Autogenerate BUILD files 2017-12-23 13:12:11 -08:00
genkubedocs Merge pull request #53631 from dixudx/enforce_cobra_required_flags 2018-01-18 02:00:27 -08:00
genman Fix scheduler refs in BUILD files. 2018-01-05 15:05:01 -08:00
genswaggertypedocs Autogenerate BUILD files 2017-12-23 13:12:11 -08:00
genutils Autogenerate BUILD files 2017-12-23 13:12:11 -08:00
genyaml Autogenerate BUILD files 2017-12-23 13:12:11 -08:00
gke-certificates-controller Autogenerate BUILD files 2017-12-23 13:12:11 -08:00
hyperkube switch hyper to cobra 2018-01-24 11:02:19 -05:00
importverifier Autogenerate BUILD files 2017-12-23 13:12:11 -08:00
kube-apiserver Merge pull request #58462 from NickrenREN/va-to-beta 2018-02-01 18:50:25 -08:00
kube-controller-manager Merge pull request #59059 from smarterclayton/move_partial_object 2018-02-05 12:40:37 -08:00
kube-proxy Merge pull request #57754 from ncdc/fix-kube-proxy-config-docs-and-json-tags 2018-02-05 19:43:42 -08:00
kube-scheduler Scheduler is not able to read from config file if configmap is not present 2018-02-05 20:39:22 -05:00
kubeadm IPv6: Ensure calculated node CIDR size for pod subnets is valid 2018-02-06 20:13:17 +00:00
kubectl switch hyper to cobra 2018-01-24 11:02:19 -05:00
kubelet Add context to all relevant cloud APIs 2018-02-06 12:49:17 -08:00
kubemark Autogenerate BUILD files 2017-12-23 13:12:11 -08:00
linkcheck Autogenerate BUILD files 2017-12-23 13:12:11 -08:00
BUILD Fix scheduler refs in BUILD files. 2018-01-05 15:05:01 -08:00
OWNERS