virtual machine weird time issue

C

chamelion

New Member
Sep 2, 2009
21
0
1
hey guys,

no idea what's happening here. have a linux vps which had been running fine. all of a sudden the clock has doubled in speed so the clock sync really goes off. the host machine is fine, clock is ok.

anyway to sync the clocks between host and guest? or at least figure out what's up with the guest??
 
chamelion said:
hey guys,

no idea what's happening here. have a linux vps which had been running fine. all of a sudden the clock has doubled in speed so the clock sync really goes off. the host machine is fine, clock is ok.

anyway to sync the clocks between host and guest? or at least figure out what's up with the guest??
Click to expand...

are you talking about a KVM guest or a container?

pls provide all details, including you Proxmox VE version (output of 'proxversion -v') and your guest OS.
 
tom said:
are you talking about a KVM guest or a container?

pls provide all details, including you Proxmox VE version (output of 'proxversion -v') and your guest OS.
Click to expand...


kvm guest fully virtualized.
proxversion -v fails, command not found. i have version 1.4. kernel 2.6.24.8-pve.

guest OS is centos 5.1 with cpanel.
 
chamelion said:
kvm guest fully virtualized.
proxversion -v fails, command not found. i have version 1.4. kernel 2.6.24.8-pve.

guest OS is centos 5.1 with cpanel.
Click to expand...

oh, sorry, the command is:

Code: 
pveversion -v
 
first, update to the latest stable proxmox ve version and see if it is better.
 
tom said:
oh, sorry, the command is:

Code: 
pveversion -v
Click to expand...

pve-manager: 1.4-7 (pve-manager/1.4/4366)
qemu-server: 1.1-5
pve-kernel: 2.6.24-16
pve-qemu-kvm: 0.11.0-2
pve-firmware: not correctly installed
vncterm: 0.9-2
vzctl: 3.0.23-1pve3
vzdump: 1.2-2
vzprocps: 2.0.11-1dso2
vzquota: 3.0.11-1dso1
 
Just upgraded to latest.

pve-manager: 1.5-1 (pve-manager/1.5/4561)
running kernel: 2.6.24-9-pve
pve-kernel-2.6.24-7-pve: 2.6.24-11
pve-kernel-2.6.24-9-pve: 2.6.24-18
pve-kernel-2.6.24-8-pve: 2.6.24-16
qemu-server: 1.1-10
pve-firmware: not correctly installed
libpve-storage-perl: 1.0-6
vncterm: 0.9-2
vzctl: 3.0.23-1pve3
vzdump: 1.2-5
vzprocps: 2.0.11-1dso2
vzquota: 3.0.11-1dso1
 
here's something interested to keep you guys engages. check "date" output on the VPS if i run it every 1-2 seconds:

root@vps [~]# date
Fri Jan 1 21:53:46 GMT 2010
root@vps [~]# date
Fri Jan 1 21:57:08 GMT 2010
root@vps [~]# date
Fri Jan 1 21:53:48 GMT 2010
root@vps [~]# date
Fri Jan 1 21:57:10 GMT 2010
root@vps [~]# date
Fri Jan 1 21:53:49 GMT 2010
root@vps [~]# date
Fri Jan 1 21:53:49 GMT 2010
root@vps [~]# date
Fri Jan 1 21:57:11 GMT 2010
root@vps [~]# date
Fri Jan 1 21:53:50 GMT 2010
root@vps [~]# date
Fri Jan 1 21:57:13 GMT 2010
root@vps [~]# date
Fri Jan 1 21:53:51 GMT 2010
root@vps [~]# date
Fri Jan 1 21:53:51 GMT 2010
Click to expand...
yep you're reading that right. it's alternating between two system times every second! the GOOD news though is that one of the two system times has now slowed down following the update; so it's on time but every second it switches to the other clock that is 3 mins behind.
 
dietmar said:
You did not update correctly - the list of packages is wrong.

see http://pve.proxmox.com/wiki/Proxmox_VE_Kernel
Click to expand...

using ovh.. i feel doing that will kill my box as they use custom kernels? unless the initramfs -tuk command will build the necessary drivers to be able to reboot into the new kernel?

dietmar said:
Please can you post your VM config?
Click to expand...

is there a file you want me to post? config just from the GUI: guest type 2.6 linux, 2 sockets, 1 core, start at boot; 1x120gb, 2x80gb drives, e1000 interface, 1256mb ram.
 
chamelion said:
is there a file you want me to post? config just from the GUI: guest type 2.6 linux, 2 sockets, 1 core, start at boot; 1x120gb, 2x80gb drives, e1000 interface, 1256mb ram.
Click to expand...

please post /etc/qemu-server/<VMID>.conf
 
How much IO load is your VM under?

I noticed my Windows 2k3 SQL 2005 test VM's OS clock fell well behind the host clock under heavy IO workload for prolonged hours. Switching from Win VirtIO on qcow2 to IDE on RAW solved the problem. Using Sysinternals Process Explorer, I saw interrupts were consuming an AVG +60% of total VM CPU when using VirtIO. The new IDE/RAW setup has much better IO performance even though I expected RedHat's VirtIO to deliver better performance than IDE. Using IDE with qcow2 was better but not near as fast as RAW.
 
dietmar said:
please post /etc/qemu-server/<VMID>.conf
Click to expand...

name: vps4
vlan0: e1000=7A:77:17:85:92:AF
bootdisk: ide0
ostype: l26
ide0: local:104/vm-104-disk-1.qcow2
memory: 1256
onboot: 1
sockets: 2
ide1: secondary:104/vm-104-disk-1.qcow2
ide2: secondary:104/vm-104-disk-2.raw
boot: cad
freeze: 0
cpuunits: 1000
acpi: 1
kvm: 1
cores: 1

Ninjix said:
How much IO load is your VM under?

I noticed my Windows 2k3 SQL 2005 test VM's OS clock fell well behind the host clock under heavy IO workload for prolonged hours. Switching from Win VirtIO on qcow2 to IDE on RAW solved the problem. Using Sysinternals Process Explorer, I saw interrupts were consuming an AVG +60% of total VM CPU when using VirtIO. The new IDE/RAW setup has much better IO performance even though I expected RedHat's VirtIO to deliver better performance than IDE. Using IDE with qcow2 was better but not near as fast as RAW.
Click to expand...

not much load on host / VM. other vps on same host has no clock issues. also note it's not just drift, which i'd be ok with (just set up ntpd and let it do its job) -- it's switching back and forth between two clock times every 1 second, whether or not ntpd is running. one of the clock times is accurate and is the current date/time, the other one is slow and is falling further and further behind, and is unchangeable - i have ntpd running now and the clock falling behind is unaffected by it.
 
You must log in or register to reply here.
Top Bottom