Search results

Long live to Proxmox :cool:

Thanks for feedback Spirit.

On very high load (stress test), I always make Virtio crash under Win2003 (with any kernel, 2.6.32 / 2.6.35 / 2.6.38) So the only working conf very very high load is Intel e1000 for me.

Thanks for the response, I will wait a few year and ask again ;) About virtio, It always crash with intensive network usage, and I tested it with 1.6 and 2.0 dll, over those 3 kernels branch. (2.6.32 / .35 / .38) Crash = no more network packet.

Hello ! Do you have any plan to include 2.6.38 kernel to pvetest repo ? I have deeply tested pure kvm performance on 2.6.32, 2.6.35 and 2.6.38, and the cpu usage for the same task is just better :) nic:e1000 kernel:2.6.32 --- WinCPU:90% ProxCPU:100% nic:e1000 kernel:2.6.35 --- WinCPU:30%...

Hey Tom, the virtio link is still available in the Quote of post #18, please clean :)

Really ? Well no problem I removed the link from my post. However I have no control over the dl.free.fr fileserver, so you should remove the link from your post too :)

You are correct :) http://packages.debian.org/fr/squeeze/firmware-linux-nonfree are only for Broadcom BCM570* (tg3)

you can fix the firmware error with this debian package here are virtio 1.2.0, please test them under official proxmox 2.6.32 / 2.6.35 and tell us.

Hello Ashes, Could you confirm you always have the problem with 2.6.35 ?

It's a good news to see the Load Average decrease with pve-kernel-2.6.32-3-pve, because even if it's a artificial mesurement, the jump was so huge between 32-1 and 32-2 that it was scary :p

I'm sorry to tell you that Load Average is worst than last official 2.6.32-2 (witch was already very bad compared to 2.6.32-1)

If you had installed the 2.6.32-1 before, the install packages should still be on your server. So, just execute the following commands: then vim /boot/grub/menu.lst and remove the 2.6.32-2 bloc at the bottom be sure to keep the kernel 2.6.32-1-pve bloc, or your server will not reboot :)

This 2.6.32-2 kernel version made my Load Average grow up from 0.2 to 0.8. I tested it on different hardware, same behaviour. I rolled back to 2.6.32-1, no more load problem. http://forum.proxmox.com/threads/4030-After-update-to-2.6.32-2-pve-load-average-increase-from-0-to-2

Okay, after testing, I confirm that there is a bug in the pve-kernel-2.6.32-2-pve. (at least on my hardware - Dell Optiplex 380 SF - Core2Duo 8400 - 4GB) Rolling back to pve-kernel-2.6.32-1-pve or pve-kernel-2.6.18-2-pve fix the KVM load average problem.

I got the same problem than you. http://forum.proxmox.com/threads/4030-After-update-to-2.6.32-2-pve-load-average-increase-from-0-to-2 Latest 2.6.32-2 kernel update made my load average grow from 0 to 2.0 :( Previous release of proxmox (1.5 - february 2010 - 2.6.32-1 kernel) did not have this...

Hello, I just updated my proxmox farm from 2.6.32-1 to 2.6.32-2 kernel (with all package available in deb http://download.proxmox.com/debian lenny pve). All was okay except my cpu load average is now between 2 and 3, instead of 0 :confused: The VM image have not changed, so it's not their...

I also would like to be able to setup the MTU of KVM network card (vmtab101i0). For example, my Internet box force me to setup the MTU with 1492 value :p best regards,

Sorry, no other idea :(

I just see you have one Dell server ! You can install and use Openmanage to change this bios setting. (see screenshot)