Description of problem:
There are two issues here:
1. The old '-rtc-td-hack' should not be used and instead we expect only the new
-rtc base=XXX,clock=YY,driftfix=slew to work for rtc time drifts.
According to Gleb it does not work properly and should be fixed.
2. We need to use -rtc clock=host as default and not use clock=vm.
If clock=vm is used we won't compensate for drifts when the VM is paused.
Do we need a bug for libvirt to update the option flag?
(In reply to comment #2)
> Do we need a bug for libvirt to update the option flag?
I did that but they said correctly that they just abstract this options and top level mgmt stack really sets it up. I opened BZs for virt-manager and rhev-m already.
Verified on qemu-kvm-0.12.1.2-2.108.el6
For the first issue:
1.start VM with -rtc base=2006-06-17T16:01:21,clock=host,driftfix=slew.
2.start VM with -rtc base=utc,clock=vm,driftfix=slew
3.start VM with -rtc base=localtime,clock=host,driftfix=none
After every step,VM can start successfully.
For the second issue.
1.start VM with -rtc base=localtime,clock=host,driftfix=slew
2.check the rtc time by in the guest using #hwclock --show
3.stop guest 3mins
4.resume guest and re-check the rtc-time.
5.Start VM with -trc base=localtiem,clock=vm,driftfix=slew
6.Repeat steps 2-4.
After step4, Guest was compensated for drifts when the VM is paused
After step6. Guest was NOT compensated for drifts when the VM is paused.
Based on above ,the issue has already been fixed.Change status to VERIFIED.
Red Hat Enterprise Linux 6.0 is now available and should resolve
the problem described in this bug report. This report is therefore being closed
with a resolution of CURRENTRELEASE. You may reopen this bug report if the
solution does not work for you.