High IRQ Load on Win2k8R2

D

DerMike

New Member
Mar 17, 2014
4
0
1
HI!

I have a Windows 2008R2 Server (full patched) with Terminal Services enabled. Running now fine for more than 1 Year without bigger Problems on an Ubuntu 12.04 LTS with KVM and libvirt. Moving the Image to a complete new Proxmoxserver results in extrem high IRQ Loads. Especially when large amounts of data are transferred over the network.

Networkcard is an e1000, Driver for Storage is virtio. Version 1.74 from http://alt.fedoraproject.org/pub/alt/virtio-win/latest/images/

High IRQ means up to 74% of CPU Time. I've got a second Machine (complete 1:1), still on Ubuntu wich never goes higher than 12%.

After some profiling with Windows Performance Analyzer i can see that there are 2 very expensive Procedures:

in ndis.sys: ndis5InterruptDpc and in ntoskrnl.exe EtwpStackWalkDpc

I've tried so step back to virtio 1.3 = no results
more CPU Power (more vcpus) = was even worse
Stop all other VMs on the host (3 Linux Boxes without noticable load) = no results
Install a complete new Window 2008R2 = out of the Box high IRQ Load :(

No Idea left...

Lg
mike

Some infos:

pveversion
pve-manager/3.1-43/1d4b0dfb (running kernel: 2.6.32-27-pve)
root@kobannode4:~# pveversion -v
proxmox-ve-2.6.32: 3.1-121 (running kernel: 2.6.32-27-pve)
pve-manager: 3.1-43 (running version: 3.1-43/1d4b0dfb)
pve-kernel-2.6.32-27-pve: 2.6.32-121
pve-kernel-2.6.32-26-pve: 2.6.32-114
lvm2: 2.02.98-pve4
clvm: 2.02.98-pve4
corosync-pve: 1.4.5-1
openais-pve: 1.1.4-3
libqb0: 0.11.1-2
redhat-cluster-pve: 3.2.0-2
resource-agents-pve: 3.9.2-4
fence-agents-pve: 4.0.5-1
pve-cluster: 3.0-12
qemu-server: 3.1-15
pve-firmware: 1.1-2
libpve-common-perl: 3.0-13
libpve-access-control: 3.0-11
libpve-storage-perl: 3.0-19
pve-libspice-server1: 0.12.4-3
vncterm: 1.1-6
vzctl: 4.0-1pve4
vzprocps: 2.0.11-2
vzquota: 3.1-2
pve-qemu-kvm: 1.7-4
ksm-control-daemon: 1.1-1
glusterfs-client: 3.4.2-1

pveperf
CPU BOGOMIPS: 110393.88
REGEX/SECOND: 1061612
HD SIZE: 275.01 GB
BUFFERED READS: 420.60 MB/sec
AVERAGE SEEK TIME: 9.31 ms
FSYNCS/SECOND: 1896.44
DNS EXT: 21.75 ms
DNS INT: 23.71 ms
 

Attachments

  • wpa.png
    wpa.png
    26.3 KB · Views: 19
DerMike said:
HI!

Yes, i tried virtio-net too - makes no real difference :(
Click to expand...

If you have in the PVE GUI "use tablet for pointer" and several cores enabled for this VM, this VM will use much CPU, please disable "use tablet for pointer" and comments us the results.

Notes:
1 - Disable/Enable "use tablet for pointer" can be do online.
2- In mode disable, the mouse of the VM will not work perfectly in VNC mode.
 
Yes, and you can also try to fix RAM size. Do not use memory ballooning.

We had high load on 2012 RDS farms with ballooning enabled, it is working far better right now with fixed size RAM.

Christophe.
 
cesarpk said:
...

2- In mode disable, the mouse of the VM will not work perfectly in VNC mode.
Click to expand...

If you install the vmmouse driver (from vmware), the mouse driver is again in sync.
 
Hi!
Thanks for all the help!

I dont use Ballooning - i assigned 20GB Ram fixed to the Server.

"use tablet for pointer" is disabled now.
Subjectively, I see no difference

I played a bit with the schedulers last night (using CFQ). The IRQ load is same as before but the System stays more responsive...
 
DerMike said:
Hi!
Thanks for all the help!

I dont use Ballooning - i assigned 20GB Ram fixed to the Server.

"use tablet for pointer" is disabled now.
Subjectively, I see no difference

I played a bit with the schedulers last night (using CFQ). The IRQ load is same as before but the System stays more responsive...
Click to expand...

Do you see interrupts only on high network load ? (if yes, on many mbit/s ? )
 
You must log in or register to reply here.
Top Bottom