Add cordon reason in log messages for cordon operations

[snay2]

Issue #, if available: Fixes #386

Description of changes:

drainEvent.Description indicates the reason for needing to cordon and/or drain a node, as well as some other details like the time the event will occur. Currently, the following reasons are possible:

• Spot ITN
• Scheduled event
• Rebalance recommendation
• SQS event, which can include ASG terminations, EC2 state change, Spot ITN, or rebalance recommendations

These changes pass the cordon reason down to each level where it may be needed for a log message but do not otherwise change the cordon/drain behavior.

An example log message, before the change:

2021/09/13 18:28:28 INF Node successfully cordoned and drained node_name=ip-192-168-0-1.ec2.internal

After:

2021/09/13 18:28:28 INF Node successfully cordoned and drained node_name=ip-192-168-0-1.ec2.internal reason="ASG Lifecycle Termination event received. Instance will be interrupted at 2021-09-13 18:28:16 +0000 UTC \n"

By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.

[bwagner5]

looks like one of the unit tests needs updated with the new function signature.

pkg/node/node_test.go:64:28: not enough arguments in call to tNode.CordonAndDrain
	have (string)
	want (string, string)

[bwagner5]

rather than using drainEvent.Kind can we use drainEvent.Description? On one hand, Kind seems like a minimal amount of information which is pretty nice. But the problem is that SQS_TERMINATION doesn't really have any meaning. When NTH is running in sqs-processor mode, all the reasons will be SQS_TERMINATION even when the SQS_TERMINATION event could actually be a SPOT_ITN, SPOT_RBR, ASG_LIFECYCLE_HOOK, etc.

wdut?

[snay2]

I agree, the SQS case specifically needs more detail. Description looks like a much better choice. Will update the PR to use that instead.

[bwagner5]

lgtm, thanks!