Description of problem: Chrony and ntpd are present in overcloud-full images . This causes an issue because chronyd starts before ntpd but is not configured thus making time synching unfunctionnal. Version-Release number of selected component (if applicable): How reproducible: Always Steps to Reproduce: 1. Deploy an overcloud 2. 3. Actual results: chronyd and ntpd are starting at boot time but ntpd fails to start because the NTP port is already binded by chronyd Expected results: Succeds Additional info:
Puppet should probably ensure that chronyd is not started, but you can do this yourself with: sudo virt-customize -a overcloud-full.qcow2 --run-command "systemctl disable chronyd"
I've checked and all 7.3 based images I have available currently have chronyd enabled by default. I check a couple 7.2 images as well and those also have it enabled by default. This applies to both OSP overcoud images and RHEL Guest Images, so it's best to assume that chronyd is always on by default in a RHEL 7 image.
Mike: Puppet doesn't manage chronyd at all so it doesn't make sense to disable the service, because we don't know exactly what is in the image. Puppet can't disable things we don't manage in the catalog.
(In reply to Emilien Macchi from comment #6) > Mike: Puppet doesn't manage chronyd at all so it doesn't make sense to > disable the service, because we don't know exactly what is in the image. > Puppet can't disable things we don't manage in the catalog. It appears that all RHEL 7 images that are publicly available have chrony installed and running by default, so something from either DIB or Puppet should be disabling chronyd if its causing a problem. Maybe it's the element that installs ntp that needs to disable chronyd.
Moving the bug to Ocata, since we can't easily backport the fix upstream I think. Omri, can we check that chrony is not in the image anymore? Thanks.
Confirmed chrony is no longer installed on OSP 11 images
Latest RC build 2017-04-28.4 [stack@undercloud-0 ~]$ rpm -q rhosp-director-images rhosp-director-images-11.0-20170428.1.el7ost.noarch [heat-admin@controller-0 ~]$ rpm -q chrony package chrony is not installed [heat-admin@controller-0 ~]$ systemctl status chornyd Unit chornyd.service could not be found. [heat-admin@controller-0 ~]$ systemctl status ntpd ● ntpd.service - Network Time Service Loaded: loaded (/usr/lib/systemd/system/ntpd.service; enabled; vendor preset: disabled) Active: active (running) since Sun 2017-04-30 11:56:02 UTC; 28min ago Main PID: 82824 (ntpd) Looks good.
As an update, I experienced an issue regarding chronyd vs. ntpd this past week on a customer PoC. In my experience the following was in place after deployment: - chronyd was enabled and active - ntpd was disabled and inactive - ntp.conf was configured with the ntp server from the deploy command - chrony.conf was not configured with the ntp server. It still had default rhel pool servers We worked around by feeding the following script in on firstboot. uname -r|grep el7 && grep ^server /etc/chrony.conf|grep rhel && cp -rp /etc/chrony.conf /etc/chrony.conf.orig && cat /etc/chrony.conf.orig|sed s/server/#server/g > /etc/chrony.conf && echo server YOURNTPSEVER iburst >> /etc/chrony.conf && systemctl restart chronyd
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory, and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://access.redhat.com/errata/RHEA-2017:1247