You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
OS: Ubuntu 22 HPC image (microsoft-dsvm:ubuntu-hpc:2204:22.04.2023111801) though likely any Ubuntu 22 image will cause this.
waagent --version
WALinuxAgent-2.9.0.4 running on ubuntu 22.04
Python: 3.10.12
Goal state agent: 2.9.1.1
VMs that change their hostname are not registering their hostname in Azure internal DNS. The logs show that the hostname change is detected, and then that eth0 link is set down and then up, however using nslookup the DNS entry is never actually updated. So WALA seems to be doing everything it intends to do, but the result is that internal DNS is never updated. Ubuntu 20 and Almalinux 8 worked without issue in the same cluster.
There is no error in the log, it is a silent bug. Setting eth0 down and up again does not cause dhcp to renew the hostname.
Serial console output is irrelevant here.
To reproduce, start an Ubuntu 22 HPC VM, run hostnamectl set-hostname SOME_TEMPORARY_HOSTNAME, wait a few minutes and run nslookup IP_ADDRESS.
As a mitigation for our users (Azure CycleCloud with Slurm), we are telling our customers to run systemctl restart systemd-networkd, which causes the hostname to be registered in internal DNS. This seems to be the behaviour that WALA does on Ubuntu 20, after reading through the source code.
The text was updated successfully, but these errors were encountered:
OS: Ubuntu 22 HPC image (microsoft-dsvm:ubuntu-hpc:2204:22.04.2023111801) though likely any Ubuntu 22 image will cause this.
waagent --version
WALinuxAgent-2.9.0.4 running on ubuntu 22.04
Python: 3.10.12
Goal state agent: 2.9.1.1
VMs that change their hostname are not registering their hostname in Azure internal DNS. The logs show that the hostname change is detected, and then that eth0 link is set down and then up, however using nslookup the DNS entry is never actually updated. So WALA seems to be doing everything it intends to do, but the result is that internal DNS is never updated. Ubuntu 20 and Almalinux 8 worked without issue in the same cluster.
hostnamectl set-hostname SOME_TEMPORARY_HOSTNAME
, wait a few minutes and run nslookup IP_ADDRESS.As a mitigation for our users (Azure CycleCloud with Slurm), we are telling our customers to run
systemctl restart systemd-networkd
, which causes the hostname to be registered in internal DNS. This seems to be the behaviour that WALA does on Ubuntu 20, after reading through the source code.The text was updated successfully, but these errors were encountered: