Support retrieving the VM UUID on Windows

[benmoss]

What type of PR is this?
/kind feature

What this PR does / why we need it:
Adds support to the vSphere cloud provider for retrieving a VM's UUID when running on Windows.

Special notes for your reviewer:
On linux VMs, there is a file with the UUID in it. On windows, this file does not exist, so we use the wmic CLI tool to get the same value.

$ wmic csproduct get UUID
UUID
5EE03942-CFD7-FE7D-A250-65899E639441

Does this PR introduce a user-facing change?:

vSphere cloud provider correctly retrieves the VM's UUID when running on Windows

[k8s-ci-robot]

Thanks for your pull request. Before we can look at your pull request, you'll need to sign a Contributor License Agreement (CLA).

📝 Please follow instructions at https://git.k8s.io/community/CLA.md#the-contributor-license-agreement to sign the CLA.

It may take a couple minutes for the CLA signature to be fully registered; after that, please reply here with a new comment and we'll verify. Thanks.

---

• If you've already signed a CLA, it's possible we don't have your GitHub username or you're using a different email address. Check your existing CLA data and verify that your email is set on your git commits.
• If you signed the CLA as a corporation, please sign in with your organization's credentials at https://identity.linuxfoundation.org/projects/cncf to be authorized.
• If you have done the above and are still having issues with the CLA being reported as unsigned, please email the CNCF helpdesk: helpdesk@rt.linuxfoundation.org

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository. I understand the commands that are listed here.

[benmoss]

/sig windows

[benmoss]

/sig vmware

[benmoss]

/check-cla

[benmoss]

/retest

[neolit123]

thank you for this PR @benmoss
commits should be probably squashed.

/retest
/priority important-longterm

[neolit123]

the wmic csproduct get UUID usage SGTM 👍

[benmoss]

@neolit123 squashed as you requested

[astrieanna]

@neolit123 what needs to happen for this to get merged?

[neolit123]

/assign @SandeepPissay

the project is fairly low activity before xmas and ny.
if the assigned reviewers and approvers do not respond try pinging them on slack at #sig-vmware and/or #sig-cloud-provider

[SandeepPissay]

@benmoss, The code changes looks good to me. Have you tested that vSphere Cloud Provider works in all cases if k8s cluster is deployed as Windows VMs on vSphere?

[benmoss]

@SandeepPissay yup, as far as we've understood the functionality everything seems to work with this.

[PatrickLang]

cc @michmike can you take a look?

[neolit123]

/assign @frapposelli @dougm

hi, can you please take a look for approval here too?
the UUID logic on Windows looks sane.

[frapposelli]

The approach is sane, I don't have the necessary Windows knowledge to validate the UUID retrieval method but

/lgtm

[dougm]

/lgtm

@dvonthenen may want to review too. We should also open an issue for porting this to the external vSphere CCM.

[dvonthenen]

/lgtm

@dvonthenen may want to review too. We should also open an issue for porting this to the external vSphere CCM.

@dougm We use the UUID reported by the kubelet. If the kubelet populates the SystemUUID property correctly, this should automatically work for Windows systems without any code changes.

[frapposelli]

/approve

[k8s-ci-robot]

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: benmoss, frapposelli

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:

• pkg/cloudprovider/providers/vsphere/OWNERS [frapposelli]

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

[benmoss]

/retest

[rhockenbury]

Having trouble getting windows nodes to find vm id with 1.14 beta.1:

E0228 18:07:38.491060    4324 datacenter.go:78] Unable to find VM by UUID. VM UUID: a7ff1742-06d1-549d-712b-193a270b7917
E0228 18:07:38.491060    4324 vsphere.go:1467] Cannot find VM runtime host. Get zone for node ul-win-node-5k error
E0228 18:07:38.491060    4324 kubelet_node_status.go:68] Unable to construct v1.Node object for kubelet: failed to get zone from cloud provider: No VM found

On linux, the /sys/class/dmi/id/product_serial file is used, which contains an id with a VMware prefix:

$ cat /sys/class/dmi/id/product_serial
VMware-42 17 76 c6 8e 65 8a 09-28 e9 1b 55 34 fa b9 76

To return the equivalent form on Windows, it looks like the following can be used:

$ wmic bios get serialnumber
SerialNumber
VMware-42 17 ff a7 d1 06 9d 54-71 2b 19 3a 27 0b 79 17

or

wmic csproduct get identifyingnumber
IdentifyingNumber
VMware-42 17 ff a7 d1 06 9d 54-71 2b 19 3a 27 0b 79 17

BUT the current implementation uses wmic csproduct get uuid which returns:

UUID
A7FF1742-06D1-549D-712B-193A270B7917

So is wmic csproduct get uuid the correct logic?

It appears the logic was correct at the time of opening the PR. However, in light of the change on the linux side from /sys/class/dmi/id/product_uuid to /sys/class/dmi/id/product_serial as implemented in #59519, the logic used on the windows side needs to be modified.

[rhockenbury]

/lgtm
@dvonthenen may want to review too. We should also open an issue for porting this to the external vSphere CCM.

@dougm We use the UUID reported by the kubelet. If the kubelet populates the SystemUUID property correctly, this should automatically work for Windows systems without any code changes.

On windows, the systemUUID does not currently get populated because of how cadvisor discovers it (https://github.com/google/cadvisor/blob/master/utils/sysfs/sysfs.go#L237).

[dvonthenen]

@rhockenbury Yes, the wmic csproduct get uuid is the correct implementation. The in-tree vSphere cloud provider code populates the UUID using that command. Please see below:
https://github.com/kubernetes/kubernetes/blob/master/pkg/cloudprovider/providers/vsphere/vsphere_util_windows.go#L28

[rhockenbury]

Yes, but this makes the assumption that the uuid and serial are the same, which is only true on some vm versions. vmware-archive#450 (comment), vmware-archive#450 (comment)

[benmoss]

Agreed, from what I remember of why I chose get uuid it was just because it required less formatting than the serial. I wasn't aware that there were instances where they disagreed. Do you want to submit a PR @rhockenbury?

[rhockenbury]

I don't have an environment set up where I could get a PR out quickly. Possibly by end of week, so if someone else is able and willing to get to it sooner, please go for it. Would this need to be in by code freeze (3/7) to make it into the 1.14 release?