New RHEL9.8 package updates clobbers /etc/chrony.conf (affects dnf, subscription-manager)
We have been upgrading RHEL 9.7 to RHEL 9.8 since May 19th. Yesterday, we lost our chrony configuration for network time protections this week going to an more recent kernel 5.14.0-687.12.1
BACKGROUND
- We have multiple RHEL 9.7 systems and been upgrading since 5/9.
- This week, starting June 1st, going from 5.14.0-611.54.1 to 5.14.0-687.12.1, we have the following issues:
- 6/01 kernel 5.14.0-687.12.1 was newer than week before (5.14.0-687.10.1)
- checking for additional packages after rebooting the server
The command ~~~ $ sudo yum check-update~~~ resulted in:
Updating Subscription Management repositories. The system clock is skewed. There is a time difference of 77.0 seconds with the entitlement server. Please check your clock settings to ensure access to all entitled content. Last metadata expiration check: 2:49:16 ago on Mon 01 Jun 2026 12:53:44 PM EDT.
** IF CRONYD IS STILL WORKING -- all these commands will work **
sudo systemctl restart chronyd sudo chronyc makestep sudo chronyc tracking
** However this is what the upgrade to RHEL 9.8 6/1 kernel results were **
$ sudo systemctl restart chronyd [sudo] password for user: Failed to restart chronyd.service: Unit chronyd.service not found. $ sudo systemctl status chronyd Unit chronyd.service could not be found. $ sudo systemctl enable --now chronyd Failed to enable unit: Unit file chronyd.service does not exist. $
Also the standard user command to check on chrony sources fails:
$ chronyc sources -v -bash: chronyc: command not found $
- SO you have all the evidence you need - RHEL 9.8 as of 6/01/2026 upgrades to RHEL 9.7 can botch your time settings*
** THE CURE **
This is short version to help you solve for missing chronyd
-
re-install package
~~~ sudo dnf install chrony ~~~-
update chrony.conf
- locate and comment pool line and
-
add your local NTP server (network core, etc.) using server directive (used 10.0.0.10 as example):
sudo vi /etc/chrony.conf ## pool 2.rhel.pool.ntp.org iburst server 10.0.0.10 iburst :wq -
If needed manually remove time skew
sudo timedatectl set-time "2026-06-01 16:12:30" -
enable and restart chronyd; then check status:
sudo systemctl enable --now chronyd sudo systemctl status chronyd -
complete checks:
sudo chronyc makestep sudo chronyc tracking sudo dnf check-update # received smaller timeskew, but still present. wait 30 seconds -
*Note this is where I added ~~~iburst~~~ at end of ~~~server~~~ line
sudo vi /etc/chrony.conf # changed "server 10.0.0.10" to "server 10.0.0.10 iburst :wq -
restart chronyd daemon and check status
sudo systemctl restart chronyd sudo chronyc sources -v ... sudo chronyc sources -v
-
** Eureka - dnf check-update worked without issue
~~~ $ sudo yum check-update Updating Subscription Management repositories. Last metadata expiration check: 0:01:55 ago on Mon 01 Jun 2026 04:16:28 PM EDT. $ ~~~
Task complete.
Hope this helps you.
** Earlier RHEL 9.8 issues (May 19th to May 26th)**
-
Previously, in May, there were no operational issues going from 5.14.0-687.5.1 to 5.14.0-687.10.1
-
minor issue since we run Microsoft SQL tools with Microsoft's unixODBC built for RedHat where /etc/yum.repos.d/redhat.repo needed line in "appstream-rpm" (exclude=unixODBC*) to prevent newer unixODBC from installing RH's own unixODBC-2.3.12 which is incompatible
~~~ $ rpm -qa | grep -i ODBC unixODBC-2.3.11-1.rh.x86_64 unixODBC-devel-2.3.11-1.rh.x86_64 msodbcsql18-18.6.2.1-1.x86_64 ~~~ -
2nd minor issue is RHEL 9.8 seems to find a weak dependency and installed bluetooth package ("bluez") which we needed to uninstall, and then modify the DNF configuration to not install weak dependency flagged packages
-
to result, I added the following lines (customize for your preferences) to /etc/dnf/dnf.conf
~~~ #// 20260519 LJT String minimalist enforcement for headless workstation install_weak_deps=False ~~~
-
-
Responses