Genererate max pods file from EC2 API

[mogren]

Description of changes:

• Based on changes in Refactor ENI limit struct aws/amazon-vpc-cni-k8s#1035. make generate-max-pods will generate this file.
• Restrictions on f1.16xlarge, g3.16xlarge, h1.16xlarge, i3.16xlarge, and r4.16xlarge have been removed from the documentation.
• Added missing metal instances, c6g and r6g.

Alphabetically sorted diff to make changes easier to see:

1c1,16
< # Mapping is calculated from AWS ENI documentation, with the following modifications:
---
> # Copyright Amazon.com Inc. or its affiliates. All Rights Reserved.
> #
> # Licensed under the Apache License, Version 2.0 (the "License"). You may
> # not use this file except in compliance with the License. A copy of the
> # License is located at
> #
> #     http://aws.amazon.com/apache2.0/
> #
> # or in the "license" file accompanying this file. This file is distributed
> # on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either
> # express or implied. See the License for the specific language governing
> # permissions and limitations under the License.
> #
> # This file was generated at 2020-07-15T13:15:46-07:00
> #
> # Mapping is calculated from AWS EC2 API using the following formula:
5c20
< # # of ENI * (# of IPv4 per ENI - 1)  + 2
---
> #   # of ENI * (# of IPv4 per ENI - 1) + 2
9,13d23
< # If f1.16xlarge, g3.16xlarge, h1.16xlarge, i3.16xlarge, and r4.16xlarge
< # instances use more than 31 IPv4 or IPv6 addresses per interface, they cannot
< # access the instance metadata, VPC DNS, and Time Sync services from the 32nd IP
< # address onwards. If access to these services is needed from all IP addresses
< # on the interface, we recommend using a maximum of 31 IP addresses per interface.
17a28
> a1.metal 234
45c56
< c5a.8xlarge 234
---
> c5a.8xlarge 234
71a83
> c5n.metal 737
72a85,93
> c6g.12xlarge 234
> c6g.16xlarge 737
> c6g.2xlarge 58
> c6g.4xlarge 234
> c6g.8xlarge 234
> c6g.large 29
> c6g.medium 8
> c6g.metal 737
> c6g.xlarge 58
79c100
< f1.16xlarge 242
---
> f1.16xlarge 394
84c105
< g3.16xlarge 452
---
> g3.16xlarge 737
95c116
< h1.16xlarge 452
---
> h1.16xlarge 737
104c125
< i3.16xlarge 452
---
> i3.16xlarge 737
116a138
> i3en.metal 737
156a179
> m5ad.16xlarge 737
159a183
> m5ad.8xlarge 234
193a218
> m6g.metal 737
207c232
< r4.16xlarge 452
---
> r4.16xlarge 737
230a256
> r5ad.16xlarge 737
233a260
> r5ad.8xlarge 234
260a288,296
> r6g.12xlarge 234
> r6g.16xlarge 737
> r6g.2xlarge 58
> r6g.4xlarge 234
> r6g.8xlarge 234
> r6g.large 29
> r6g.medium 8
> r6g.metal 737
> r6g.xlarge 58
283a320,321
> u-18tb1.metal 737
> u-24tb1.metal 737

By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.

[saiteja313]

👍

[pysysops]

Does this account for custom ENIConfig MAX_PODS calculations? It'd be great if it could tackle that frustrating bug.

[mogren]

@pysysops unfortunately not, since custom networking is a config setting on the CNI, and --max-pods is a flag on the kubelet itself. We are working on other solutions to this problem, but right now overriding the flag in bootstrap arguments is the only solution.