Ask Your Question
1

live-migration not working with slightly different CPU

asked 2013-09-06 11:16:40 -0500

jsfrerot gravatar image

updated 2013-10-08 00:09:22 -0500

fifieldt gravatar image

Hi, I've encountered a weird problem with live migration. Let me give you some history.

I have 2 compute nodes, compute01 and compute02 running grizzly under ubuntu 13.04 compute01 has this cpu:

model name : Intel(R) Core(TM) i3-3220 CPU @ 3.30GHz

flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx rdtscp lm constant_tsc arch_perfmon pebs bts rep_good nopl xtopology nonstop_tsc aperfmperf eagerfpu pni pclmulqdq dtes64 monitor ds_cpl vmx est tm2 ssse3 cx16 xtpr pdcm pcid sse4_1 sse4_2 popcnt tsc_deadline_timer xsave avx f16c lahf_lm arat epb xsaveopt pln pts dtherm tpr_shadow vnmi flexpriority ept vpid fsgsbase smep erms

and compute02:

model name : Intel(R) Core(TM) i5-3470 CPU @ 3.20GHz

flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx rdtscp lm constant_tsc arch_perfmon pebs bts rep_good nopl xtopology nonstop_tsc aperfmperf eagerfpu pni pclmulqdq dtes64 monitor ds_cpl vmx smx est tm2 ssse3 cx16 xtpr pdcm pcid sse4_1 sse4_2 x2apic popcnt tsc_deadline_timer aes xsave avx f16c rdrand lahf_lm ida arat epb xsaveopt pln pts dtherm tpr_shadow vnmi flexpriority ept vpid fsgsbase smep erms

so slightly different (ok, slightly can be very subjective !)

anyhow, So I started a VM with 4G ram and 2Vcpu and it started on compute01 then I live-migrated to compute02 without problems, I was very happy, but then I wanted to migrate it back to compute01 and got errors: http://paste.openstack.org/show/45928/

After searching the net I found this thread which help me to migrate my server back to compute01: https://ask.openstack.org/en/question/4548/live-migration-invalidcpuinfo-unacceptable-cpu-info-cpu-doesnt-have-compatibility/

I actually commented out this line in /usr/share/pyshared/nova/virt/libvirt/driver.py self._compare_cpu(source_cpu_info)

I know it's not really fixing the source of the issue, but at least I can live migrate...

Is there any change we can get that really fixed ?

I'm available for debugging if needed.

edit retag flag offensive close merge delete

1 answer

Sort by ยป oldest newest most voted
0

answered 2013-09-07 00:12:47 -0500

fetahi gravatar image

You need to specify the appropriate libvirt_cpu_mode and libvirt_cpu_model flags in nova.conf. I believe the default is to use the host CPU directly and hence all flags are checked.

edit flag offensive delete link more

Comments

jsfrerot gravatar imagejsfrerot ( 2013-09-07 10:10:38 -0500 )edit

I got live migration work by setting this two lines in nova.conf on each computing node: live_migration_flag=VIR_MIGRATE_UNDEFINE_SOURCE,VIR_MIGRATE_PEER2PEER,VIR_MIGRATE_LIVE libvirt_cpu_mode=none

shedoh gravatar imageshedoh ( 2013-09-09 04:48:32 -0500 )edit

This is not working for me either.

jsfrerot gravatar imagejsfrerot ( 2013-09-17 07:39:31 -0500 )edit

Your Answer

Please start posting anonymously - your entry will be published after you log in or create a new account.

Add Answer

Get to know Ask OpenStack

Resources for moderators

Question Tools

Stats

Asked: 2013-09-06 11:16:40 -0500

Seen: 725 times

Last updated: Sep 07 '13