Open bwilkinscloud opened 1 year ago
I updated the title to now reflect the actual problem and provided the fix.
The bad state part was due to cloud-init. That problem is covered under: https://bugzilla.redhat.com/show_bug.cgi?id=1957532
and the workaround for RHEL8.7 because it does not have the fixed version of cloud-init is: For VMs that were upgraded from RHEL 8.4 or earlier, you must work around the issue manually.
To do so, edit the cloud.cfg
file and changing the ssh_genkeytypes: ~
line to ssh_genkeytypes: ['rsa', 'ecdsa', 'ed25519']
. This makes it possible for SSH keys to be deleted and generated correctly when provisioning a RHEL 8 VM in the described circumstances.
I recommend you put the nmcli and the cloud.cfg into an Ansible playbook then run it from your hardening pipeline that should have an automated capture capability (e.g. with terraform)
Running sudo waagent -deprovision+user and I got this error:
ERROR MainThread Command: [nmcli device modify eth0 ipv4.dhcp-hostname localhost.localdomain ipv6.dhcp-hostname localhost.localdomain], return code: [6], result: [Error: Reapplying connection to device 'eth0' (/org/freedesktop/NetworkManager/Devices/2) failed: The settings specified are invalid: ipv6.addr-gen-mode: property is invalid
The fix is to do the following:
Run
$ nmcli device modify eth0 ipv4.dhcp-hostname localhost.localdomain ipv6.dhcp-hostname localhost.localdomain ipv6.addr-gen-mode eui64
before
running sudo waagent deprovision
Distro: RHEL8.7 WALinuxAgent version: 2.7.0.6 running on redhat 8.7 Python: 3.6.8
Additional context N/A
Log file attached N/A