live migration didn´t work on AMD?

[valshare]

Hello,

i have the problem, that live migration between my AMD Servers didn´t work with windows kvms. As a shared storage i use openfile with nfs support. The openfile has all updates installed. If i migrate the kvm to the other server, the migration finished but the machine freez and i must stop the machine.

I have installed a new machine and checked the follow:

• with and without network

• enabled and disabled kvm

• enabled and disabled acpi

• single and multi cpu/core for the kvm

When i migrated the machine, if dos is booted in the maintenance menu, migration works without problem. But if windows is startet, kvm machine freez when i migrate it.

Nothing helped. Is live migration working for AMD? Does anyone has working it on AMD CPU?

proxmox01:~# pveversion -v
pve-manager: 1.5-7 (pve-manager/1.5/4660)
running kernel: 2.6.24-10-pve
proxmox-ve-2.6.24: 1.5-21
pve-kernel-2.6.24-7-pve: 2.6.24-11
pve-kernel-2.6.24-10-pve: 2.6.24-21
qemu-server: 1.1-11
pve-firmware: 1.0-3
libpve-storage-perl: 1.0-10
vncterm: 0.9-2
vzctl: 3.0.23-1pve8
vzdump: 1.2-5
vzprocps: 2.0.11-1dso2
vzquota: 3.0.11-1
pve-qemu-kvm: 0.11.1-2

processor	: 7
vendor_id	: AuthenticAMD
cpu family	: 16
model		: 4
model name	: Quad-Core AMD Opteron(tm) Processor 2376
stepping	: 2
cpu MHz		: 2300.092
cache size	: 512 KB
physical id	: 1
siblings	: 4
core id		: 3
cpu cores	: 4
fpu		: yes
fpu_exception	: yes
cpuid level	: 5
wp		: yes
flags		: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt pdpe1gb rdtscp lm 3dnowext 3dnow constant_tsc rep_good pni cx16 popcnt lahf_lm cmp_legacy svm extapic cr8_legacy abm sse4a misalignsse 3dnowprefetch osvw ibs skinit wdt
bogomips	: 4600.20
TLB size	: 1024 4K pages
clflush size	: 64
cache_alignment	: 64
address sizes	: 48 bits physical, 48 bits virtual
power management: ts ttp tm stc 100mhzsteps hwpstate

 

[pille]

i've got the same symptoms here, but my machines are running on Intel Xeon.

summary:
* KVM (kernel 2.6.32-1-pve)
* shared storage on NFS
* VM is ubuntu 9.10
* net is VIRTIO
* offline migration works well
* after online migration:
* VM has high CPU load (being idle)
* VNC is available, but input is buggy
* ping unreliable
* SSH connection refused

will try out 2.6.18...

 

[valshare]

Hi Pille,

exakt the same problems that you write down here on AMD.

Regards, Valle

 

[pille]

ok, 2.6.18-2-pve works fine without these issues here.
even my ssh session kept alive while migrating.

another observation i made on kernel 2.6.32-1-pve:
suddenly the pings came back (after a long time)
but still no ssh:

64 bytes from 192.168.2.21: icmp_seq=1089 ttl=63 time=0.911 ms
64 bytes from 192.168.2.21: icmp_seq=1090 ttl=63 time=0.936 ms
64 bytes from 192.168.2.21: icmp_seq=1091 ttl=63 time=0.717 ms
64 bytes from 192.168.2.21: icmp_seq=1092 ttl=63 time=0.714 ms

64 bytes from 192.168.2.21: icmp_seq=1093 ttl=63 time=1367883 ms
64 bytes from 192.168.2.21: icmp_seq=1094 ttl=63 time=1366883 ms
64 bytes from 192.168.2.21: icmp_seq=1095 ttl=63 time=1365884 ms
64 bytes from 192.168.2.21: icmp_seq=1096 ttl=63 time=1364884 ms
64 bytes from 192.168.2.21: icmp_seq=1097 ttl=63 time=1363885 ms
64 bytes from 192.168.2.21: icmp_seq=1098 ttl=63 time=1362885 ms
64 bytes from 192.168.2.21: icmp_seq=2461 ttl=63 time=0.963 ms
64 bytes from 192.168.2.21: icmp_seq=2462 ttl=63 time=0.695 ms

 

[pille]

dear proxmox team,

can you reproduce this bug in your testbed?

 

[pille]

hey proxmox team,

if you plan on releasing PVE-2.0 with kernel 2.6.32, you should check for this bug.

 

[valshare]

Hi Pille,

live migration works now with Kernel 2.6.24 but didn´t work with 2.6.18 for me.
But Kernel 2.6.24 i can´t use because there is a time drifting bug. In one minute the clock drift -5 seconds in Windows XP. We use time critical application that needs a clock that works without a drifting issue.

Regards, Valle

 

[Robstarusa]

I'm having a similar issue.

Both 2.6.32 boxes.

Migrating from an amd 5050e to an athlon ii x4 620. on iscsi/lvm.

This is as far as it gets:

/usr/sbin/qmigrate --online 192.168.75.86 102
Apr 06 17:20:41 starting migration of VM 102 to host '192.168.75.86'
Apr 06 17:20:41 copying disk images
Logging in to [iface: default, target: iqn.2010-03.us.gurnee.vm3.istgt0:disk1, portal: 192.168.75.9,3260]
Login to [iface: default, target: iqn.2010-03.us.gurnee.vm3.istgt0:disk1, portal: 192.168.75.9,3260]: successful
Logging in to [iface: default, target: iqn.2010-03.us.gurnee.vm3.istgt0:disk0, portal: 192.168.75.9,3260]
Login to [iface: default, target: iqn.2010-03.us.gurnee.vm3.istgt0:disk0, portal: 192.168.75.9,3260]: successful
Logging in to [iface: default, target: iqn.2010-03.us.gurnee.vm3.istgt0:disk2, portal: 192.168.75.9,3260]
Login to [iface: default, target: iqn.2010-03.us.gurnee.vm3.istgt0:disk2, portal: 192.168.75.9,3260]: successful
Apr 06 17:20:49 starting VM on remote host '192.168.75.86'
Apr 06 17:20:49 starting migration tunnel
Apr 06 17:20:49 starting online/live migration
Apr 06 17:20:51 migration status: active (transferred 4396KB, remaining 536532KB), total 540928KB)

It just stops here....it doesn't succeed, doesn't fail.

Both boxes have 8GB ram and are running the same kernel with the same updates. Guest is KVM with 1 cpu, 512 ram and 8G "disk".

 

[ashes]

I have problem with live migration. My CPU is Intel Xeon. My storage is NFS.

pve-manager: 1.5-8 (pve-manager/1.5/4674)
running kernel: 2.6.32-1-pve
proxmox-ve-2.6.32: 1.5-4
pve-kernel-2.6.32-1-pve: 2.6.32-4
pve-kernel-2.6.18-1-pve: 2.6.18-4
qemu-server: 1.1-11
pve-firmware: 1.0-3
libpve-storage-perl: 1.0-10
vncterm: 0.9-2
vzctl: 3.0.23-1pve8
vzdump: 1.2-5
vzprocps: 2.0.11-1dso2
vzquota: 3.0.11-1
pve-qemu-kvm: 0.11.1-2
ksm-control-daemon: 1.0-3