Fix EC2MetadataInstanceInfo ENI calculation

[AndrewSirenko]

Is this a bug fix or adding new feature?
bug fix

What is this PR about? / Why do we need it?
Fixes #2064; Fixes a regression in calculating currently attached ENIs in our EC2MetadataInstanceInfo function to calculate CSINode allocatable count when not using reservedVolumeAttachments.

Instead of checking for newlines, we should check for counts of a relevant regex, to avoid inconsistencies when IMDS returns or doesn't return a newline at end of response.

What testing is done?
Can confirm that there is no newline when querying IMDS for number of attached ENIs:

> && curl -H "X-aws-ec2-metadata-token: $TOKEN" http://169.254.169.254/latest/meta-data/network/interfaces/macs
  % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
                                 Dload  Upload   Total   Spent    Left  Speed
100    56  100    56    0     0  46128      0 --:--:-- --:--:-- --:--:-- 56000
0e:1c:7d:81:2b:19/
0e:8c:22:a2:16:ef/sh-4.2$

MAC address regex is sufficient across Linux and Windows CSI Driver deploys.

Open Question

Should we be smarter here and regex for a mac address instead of newlines? Seems like other IMDS endpoints DO end with a /n... Yes, new commit incoming.

[k8s-ci-robot]

Skipping CI for Draft Pull Request.
If you want CI signal for your change, please convert it to an actual PR.
You can still manually trigger a test run with /test all

[github-actions]

Code Coverage Diff

File	Old Coverage	New Coverage	Delta
github.com/kubernetes-sigs/aws-ebs-csi-driver/pkg/cloud/metadata/ec2.go	94.3%	94.4%	0.1
github.com/kubernetes-sigs/aws-ebs-csi-driver/pkg/util/util.go	62.3%	63.6%	1.4

[AndrewSirenko]

/retest

[ConnorJC3]

/retest

[AndrewSirenko]

Looking like windows failure isn't flake... Will go debug and add unit test for windows case (and perhaps snow device and metal instance cases?)

Worst case we have old version of this PR for release to at least fix regression.

[AndrewSirenko]

/retest

[AndrewSirenko]

#2063 also failing upon a retest of windows job, so looks like not related to current PR. Will investigate during working hours. A quick google of errors shows a possible certs issue on test cluster? But may be red herring. EKS remote error: tls: internal error when running kubectl logs command | AWS re:Post

Manually testing IMDS query on windows 2019 and 2022 shows that regex would work correctly.

PS C:\C\2b63e1adda895b3754e1fac883a865627c50de7743be651596b1cfb9e02283d7> Invoke-RestMethod -Headers @{"X-aws-ec2-metadata-token" = $token} -Method GET -Uri http://169.254.169.254/latest/meta-data/network/interfaces/macs
02:17:f1:05:9d:33/
PS C:\C\2b63e1adda895b3754e1fac883a865627c50de7743be651596b1cfb9e02283d7>

[AndrewSirenko]

Looks like #2066 is passing

[ConnorJC3]

/retest

[torredil]

/retest

[torredil]

Thanks for the quick turnaround on the fix, the regex approach is much more resilient than the current new line tech.

Manually tested changes:

• Windows node with 2 ENIs:

kubectl logs ebs-csi-node-windows-z2xw2 -n kube-system                                                                                               
Defaulted container "ebs-plugin" out of: ebs-plugin, node-driver-registrar, liveness-probe
I0618 18:36:19.041526    3192 main.go:157] "Initializing metadata"
I0618 18:36:19.288941    3192 ec2.go:92] "Number of attached ENIs" attachedENIs=2 Enis=<
        0a:03:b0:13:df:8b/
        0a:30:70:82:cf:6f/
 >

• Linux node with 1 ENI:

kubectl logs ebs-csi-node-bwfv5 -n kube-system                                                                                                       
Defaulted container "ebs-plugin" out of: ebs-plugin, node-driver-registrar, liveness-probe
I0618 18:36:14.821441       1 main.go:157] "Initializing metadata"
I0618 18:36:14.824222       1 ec2.go:92] "Number of attached ENIs" attachedENIs=1 Enis="0a:fa:d3:89:fa:b3/"

[ConnorJC3]

/retest

[AndrewSirenko]

/test pull-aws-ebs-csi-driver-test-e2e-external-eks-windows

[ConnorJC3]

/lgtm

[torredil]

/retest

[torredil]

/lgtm
/approve

[k8s-ci-robot]

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: torredil

The full list of commands accepted by this bot can be found here.

The pull request process is described here

Needs approval from an approver in each of these files:

• OWNERS [torredil]

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment