helm-charts: Add selinuxMount flag to enable/disable /etc/selinux host mount

[Aste88]

Describe what this PR does

Add selinuxMount flag to enable/disable /etc/selinux host mount inside pods to support selinux-enabled filesystems

Is the change backward compatible?

Yes, selinuxMount default is true (the previous behaviour)

Are there concerns around backward compatibility?

None I'm aware of

Related issues

Fixes: #2876

[Madhu-1]

@Aste88 Thanks for the PR. can you fix DCO and commit lint problem. please refer to https://github.com/ceph/ceph-csi/blob/devel/docs/development-guide.md#code-contribution-workflow for any help

[humblec]

@Aste88 @Madhu-1 @Rakshith-R @nixpanic I doubt this was the right fix we need , That said, here we are defaulting to true and consuming it from the host. If the PODs are running in host which does not support, this could cause issues? or did I miss anything in helm deployment here?

[Madhu-1]

@humblec this was done to support the backward compatibility and now the user is having an option to turn it on and off which was not present at all.

[humblec]

@humblec this was done to support the backward compatibility and now the user is having an option to turn it on and off which was not present at all.

Getting a knob done is fine, but a couple of things , like

1. Default to `true
2. HostPath

All new deployments will try to pick the hostmount now being this value default to true and if we are on NON selinux system it could cause issues.

Instead of this, I was thinking we could have gone ahead with DirectoryOrCreate instead of HostPath.. I believe its the better solution . Please let me know what you folks think.. I can drop a PR accordingly.

[Madhu-1]

@humblec this was done to support the backward compatibility and now the user is having an option to turn it on and off which was not present at all.

Getting a knob done is fine, but a couple of things , like

1. Default to `true
2. HostPath

All new deployments will try to pick the hostmount now being this value default to true and if we are on NON selinux system it could cause issues.

Yes correctly its the existing behaviour for both helm and deployments with kube templates.

Instead of this, I was thinking we could have gone ahead with DirectoryOrCreate instead of HostPath.. I believe its the better solution . Please let me know what you folks think. I can drop a PR accordingly.

Is it allowed to create a Directory on /etc folder on the nodes? If possible looks good.