vmware tools fail in Ubuntu 24.04 cloud image because `hwclock` doesn't exist
Describe the bug
This bug has been reported here https://bugs.launchpad.net/ubuntu/+source/open-vm-tools/+bug/2039206
At this point I gather an update for open-vm-tools should solve this, but this has happened yet, as tested with ubuntu 24.04 cloud image from 10 July 2024.
In my case this probably results in the network interface not being attached after powering on the virtual machine, even if the original template does have the option checked.
Reproduction steps
- Import an ova cloudimage template from here https://cloud-images.ubuntu.com/daily/server/noble/current/noble-server-cloudimg-amd64.ova
- clone the template using Packer v1.11.1 using:
vapp {
properties = {
hostname = "${var.playbook}"
user-data = base64encode(var.user_data)
}
}
[...]
customize {
linux_options {
host_name = "${var.template_name}"
domain = "company.internal"
hw_clock_utc = true
}
network_interface {
ipv4_address = "${var.template_ip}"
ipv4_netmask = "${var.vm_netmask}"
}
dns_server_list = [ "10.88.88.21", "10.88.88.22" ]
ipv4_gateway = "${var.vm_gateway}"
}
Expected behavior
That vmtools stop using hwclock and use timedatectl.
Additional context
I would like to know if there's been any progress here and if this is discussed anywhere. I've looked at the issues here and haven't found any mentions related to this topic, but I might have missed it.
The Linux guest customization Perl scripts come from the vCenter Server, not from open-vm-tools.
The scripts have been updated to skip attempting setting the hardware clock if the hwclock command is not available on the system.
The updated scripts are in the recent releases of vCenter:
- VMware vCenter Server 8.0U3 & 8.0U3a
- VMware vCenter Server 7.0 Update 3q
I understand. I had no idea vcenter itself provided the perl scripts. Thanks for the information. I just need to upgrade the vcenter server then.
Sounds like a plan.
I am pinging the Guest CustomizationTeam to:
- provide an update to the Ubuntu bug report.
- remember to note the fix is the relative Release Notes.
Just in case someone come across this thread (I know this is becoming even more off-topic, but this might help others too), you might encounter this bug: https://knowledge.broadcom.com/external/article/371524/patching-vcenter-server-to-80-u3-fails-a.html
Changing the role names as follows:
CNS-Datastore -> CNS-Vanilla-Datastore
CNS-SEARCH-AND-SPBM -> CNS-Vanilla-SEARCH-AND-SPBM
before updating (or after rolling back the failed update) works.