Various symptoms might be caused by the same issue:
Time inside virtual machine goes faster (or slower) than on the host node.
Inside of Virtual Machine you may see kernel is reporting soft lockup for some processes:
... kernel: BUG: soft lockup - CPU#0 stuck for 67s! [httpd:15051] ...
Meanwhile, in Virtual Machine's log (VM's personal log, located in
... Skip all remaining time=60000 msec!!! Skip all remaining time=60000 msec!!! ...
- A virtual machine may hang and be inaccessible for certain periods of time (2-3 minutes) and continue working normally afterwards.
The issue could be caused by firmware issue, high CPU resources over-commitment or specifics of OS behavior inside the virtual machine.
Verify that the guest OS has been optimized according to timekeeping best practices:
Install all available firmware updates for the server hardware
Disable EIST (Enhanced Intel SpeedStep® Technology) in BIOS settings.
Disable cpuspeed service on a hardware node:
# service cpuspeed stop # chkconfig cpuspeed off
Make sure that the 'Guest OS Type' is indicated correctly for the VM. For Operating systems, based on Red Hat kernel (e.g. Fedora, Cloudlinux, etc.) it is necessary to set 'Red Hat Enterprise Linux' OS Type (not 'Other Linux').
Disable VM's time synchronization with the host. Open VM's configuration file (by default
/var/parallels/$vm_name/config.pvs) for editing and locate the following section:
<TimeSync SyncInterval_patch="1" dyn_lists=""> <Enabled>1</Enabled> <SyncInterval>60</SyncInterval> <KeepTimeDiff>0</KeepTimeDiff> <SyncHostToGuest>0</SyncHostToGuest> </TimeSync>
Change "Enabled" to 0.
After the modifications, VM should be stopped and started (not restarted!):
# prlctl stop $vm_name && prlctl start $vm_name