migration and ntp

[mir]

Hi all,

I just want to hear how others handles the problem with ntp and online migration in relation to monitoring like nagios, icinga etc.
Every time an online migration is occurring ntp looses sync with the remote ntp-server and will stay out of sync for some time so I wonder if there is some magic try to prevent this problem?

 

[udo]

Hi all,

I just want to hear how others handles the problem with ntp and online migration in relation to monitoring like nagios, icinga etc.
Every time an online migration is occurring ntp looses sync with the remote ntp-server and will stay out of sync for some time so I wonder if there is some magic try to prevent this problem?

Hi,
which ntp do you mean? Host (source/destination) or inside VM?

If you talk about an ntp inside an VM - I don't use an ntp-client inside an VM (execpt I do an mistake). The correct time came from the host.

Udo

 

[mir]

Yes, I talk of inside the VM.

Is kvm-clock from /sys/devices/system/clocksource/clocksource0/current_clocksource reliable?
And does kvm-clock work properly on any platform?

 

[udo]

Hi mir,
AFAIK you got less trouble without ntp in the client as with ntp.

If that work on any platform? Do you mean host-platform or VMs (linux/Windows/xxx).
I have only some different AMD and Intelboxes and no trouble with the time. Sometimes after restarting an host which is also ceph-monitor (which is on an pve-host) ceph mean the time is not good enough. But this disapear after an short time - I think the ntp starts before the vlan-network is finished...

Udo

 

[mir]

By platform I was referring to platform in VM.
Reading this thread seems to suggest that you should use ntp even in guests: http://thread.gmane.org/gmane.comp.emulators.kvm.devel/96835
But apparently Suse and Ubuntu disagrees!
Very confusing indeed!!!

 

[udo]

By platform I was referring to platform in VM.
Reading this thread seems to suggest that you should use ntp even in guests: http://thread.gmane.org/gmane.comp.emulators.kvm.devel/96835
But apparently Suse and Ubuntu disagrees!
Very confusing indeed!!!

Hi mir,
perhaps you have less trouble with an cronjob for ntpdate inside the guest?


Udo

 

[mir]

But ntpdate is officially declared deprecated so this is not an option for the future.

" -q Exit the ntpd just after the first time the clock is set. This behavior mimics that of the ntpdate program, which is to be retired. The -g and -x options can be
used with this option. Note: The kernel time discipline is disabled with this
option.
"

 

[adoII]

Hi,

yesterday I noticed the same problem.
ntpd on vms looses ntp synchronsation when migrating virtual machines. It seems that ntp is hanging for very long time. I had to restart ntp daemons in the client.
I have ntp in all proxmox servers and linux vms.
Time moving backwards gave me lots of problems during a mail migration.

What is the recommended way to have stable time in vms also if you do migrations ?

 

[ghusson]

Normaly its makes no sense to run an ntpd inside an VM. Use the correct time at the host and all the VMs have the right time.