Ubuntu update on Azure VMs causing DNS problems

• The issue affects customers running Ubuntu 18.04 VMs who have Ubuntu Unattended-Upgrades enabled and received systemd version 237-3ubuntu10.54.
• Customers can consider rebooting impacted VM instances as a workaround so that they receive a fresh DHCP lease and new DNS resolvers.
• According to the Microsoft’s announcement, Azure Container Apps, Azure Database for PostgreSQL Azure Kubernetes Service and Azure VMware Solution are affected.

Microsoft confirmed that Azure virtual machine customers who are running Ubuntu 18.04 experiencing problems with resolving DNS. According to the announcement, the problem affects users who are running Ubuntu 18.04 virtual machines that are recently upgraded to systemd version 237-3ubuntu10.54. Canonical also confirmed the bug for Ubuntu Systemd upgrades but the announcement doesn’t mention Azure VMs. The offending package Ubuntu updates have been removed until Microsoft completes its investigation.

Systemd version 237-3ubuntu10.54

Microsoft stated that the issue started at approximately 06:00 UTC on 30 Aug 2022. Microsoft advised users to take a look at Canonical’s announcement, where a potential fix has been highlighted. Microsoft also stated that rebooting the impacted VM instances could be a potential workaround because they can receive a fresh DHCP lease and new DNS resolvers. The issue has affected various Azure services, including Azure Container Apps, Azure Database for PostgreSQL Azure Kubernetes Service, and Azure VMware Solution.

We are aware of an ongoing incident with VMs that recently upgraded to system version 237-3ubuntu 10.54 experiencing DNS error. Please keep updated by following the Azure status page here: https://t.co/cxAjC3RkIi ^CR

— Azure Support (@AzureSupport) August 30, 2022

Canonical stated that it is the same bug that they experienced on Ubuntu 20.04 two years ago. It was triggered by open-vm-tools issuing “udevadm trigger”. Canonical also advised users to reboot their instances or issue “udevadm trigger -cadd -yeth0 && systemctl restart systemd-networkd” as root. To fix a broken instance, users can run:

$ sudo udevadm trigger -cadd -yeth0 && sudo systemctl restart systemd-networkd

and install the test packages.