HE vm fails to migrate due to host CPU incompatibility even after changing cluster CPU type
Hi guys, I've started an oVirt 4.1 HE deployment on a Broadwell based server. Then I added to HE a second, older host based on Nehalem. I've downgraded the cluster CPU type to Nehalem to accommodate host2 and it finally reached score 3400. However, when I try to migrate HE vm it fails with the following error: 2017-03-03 20:19:51,814 ERROR (migsrc/b0d38435) [virt.vm] (vmId='b0d38435-5774-4ca9-ad24-70b57b5bc25d') unsupported configuration: guest and host CPU are not compatible: Host CPU does not provide required features: pclmuldq, fma, pcid, x2apic, movbe, tsc-deadline, aes, xsave, avx, fsgsbase, bmi1, hle, avx2, smep, bmi2, erms, invpcid, rtm, rdseed, adx, smap, 3dnowprefetch; try using 'Broadwell-noTSX' CPU model (migration:265) I believe the problem is in the HE vm XML where the cpu is still configured as Broadwell. how can I change this specific setting without losing the deployment? Please find all the relevant logs at the following link: https://www.dropbox.com/sh/njl9aofhdw10ses/AADf2Ql4GKVIKcbgLivbmjC2a Besides that, I believe this is a wrong behavior because HE should follow cluster properties (otherwise do not reach score 3400); do you believe is it worth opening a issue on bugzilla? Thank you, Stefano.
On Sat, Mar 4, 2017 at 12:04 PM, Stefano Stagnaro < stefanos@prismatelecomtesting.com> wrote:
Hi guys,
I've started an oVirt 4.1 HE deployment on a Broadwell based server. Then I added to HE a second, older host based on Nehalem. I've downgraded the cluster CPU type to Nehalem to accommodate host2 and it finally reached score 3400. However, when I try to migrate HE vm it fails with the following error:
2017-03-03 20:19:51,814 ERROR (migsrc/b0d38435) [virt.vm] (vmId='b0d38435-5774-4ca9-ad24-70b57b5bc25d') unsupported configuration: guest and host CPU are not compatible: Host CPU does not provide required features: pclmuldq, fma, pcid, x2apic, movbe, tsc-deadline, aes, xsave, avx, fsgsbase, bmi1, hle, avx2, smep, bmi2, erms, invpcid, rtm, rdseed, adx, smap, 3dnowprefetch; try using 'Broadwell-noTSX' CPU model (migration:265)
I believe the problem is in the HE vm XML where the cpu is still configured as Broadwell. how can I change this specific setting without losing the deployment? Please find all the relevant logs at the following link: https://www.dropbox.com/sh/njl9aofhdw10ses/AADf2Ql4GKVIKcbgLivbmjC2a
Besides that, I believe this is a wrong behavior because HE should follow cluster properties (otherwise do not reach score 3400); do you believe is it worth opening a issue on bugzilla?
I would consider opening a BZ to track this. Adding some people who may have some insight on the issue.
Thank you, Stefano.
_______________________________________________ Users mailing list Users@ovirt.org http://lists.ovirt.org/mailman/listinfo/users
-- Sandro Bonazzola Better technology. Faster innovation. Powered by community collaboration. See how it works at redhat.com
On Wed, Mar 8, 2017 at 1:02 PM, Sandro Bonazzola <sbonazzo@redhat.com> wrote:
On Sat, Mar 4, 2017 at 12:04 PM, Stefano Stagnaro <stefanos@ prismatelecomtesting.com> wrote:
Hi guys,
I've started an oVirt 4.1 HE deployment on a Broadwell based server. Then I added to HE a second, older host based on Nehalem. I've downgraded the cluster CPU type to Nehalem to accommodate host2 and it finally reached score 3400. However, when I try to migrate HE vm it fails with the following error:
2017-03-03 20:19:51,814 ERROR (migsrc/b0d38435) [virt.vm] (vmId='b0d38435-5774-4ca9-ad24-70b57b5bc25d') unsupported configuration: guest and host CPU are not compatible: Host CPU does not provide required features: pclmuldq, fma, pcid, x2apic, movbe, tsc-deadline, aes, xsave, avx, fsgsbase, bmi1, hle, avx2, smep, bmi2, erms, invpcid, rtm, rdseed, adx, smap, 3dnowprefetch; try using 'Broadwell-noTSX' CPU model (migration:265)
I believe the problem is in the HE vm XML where the cpu is still configured as Broadwell. how can I change this specific setting without losing the deployment? Please find all the relevant logs at the following link: https://www.dropbox.com/sh/njl9aofhdw10ses/AADf2Ql4GKVIKcbgL ivbmjC2a
Besides that, I believe this is a wrong behavior because HE should follow cluster properties (otherwise do not reach score 3400); do you believe is it worth opening a issue on bugzilla?
I would consider opening a BZ to track this. Adding some people who may have some insight on the issue.
The definition for the engine VM is getting extracted by ovirt-ha-agent from the OVF_STORE volume, not sure why the engine doesn't update it when you change cluster properties. I think it's work to fill a bug. Stefano, did you tried simply changing the number of cores for the engine VM from the engine to force a configuration update?
Thank you, Stefano.
_______________________________________________ Users mailing list Users@ovirt.org http://lists.ovirt.org/mailman/listinfo/users
-- Sandro Bonazzola Better technology. Faster innovation. Powered by community collaboration. See how it works at redhat.com
On mer, 2017-03-08 at 13:58 +0100, Simone Tiraboschi wrote:
On Wed, Mar 8, 2017 at 1:02 PM, Sandro Bonazzola <sbonazzo@redhat.com> wrote:
On Sat, Mar 4, 2017 at 12:04 PM, Stefano Stagnaro <stefanos@prismatelecomtesting.com> wrote: Hi guys,
I've started an oVirt 4.1 HE deployment on a Broadwell based server. Then I added to HE a second, older host based on Nehalem. I've downgraded the cluster CPU type to Nehalem to accommodate host2 and it finally reached score 3400. However, when I try to migrate HE vm it fails with the following error:
2017-03-03 20:19:51,814 ERROR (migsrc/b0d38435) [virt.vm] (vmId='b0d38435-5774-4ca9-ad24-70b57b5bc25d') unsupported configuration: guest and host CPU are not compatible: Host CPU does not provide required features: pclmuldq, fma, pcid, x2apic, movbe, tsc-deadline, aes, xsave, avx, fsgsbase, bmi1, hle, avx2, smep, bmi2, erms, invpcid, rtm, rdseed, adx, smap, 3dnowprefetch; try using 'Broadwell-noTSX' CPU model (migration:265)
I believe the problem is in the HE vm XML where the cpu is still configured as Broadwell. how can I change this specific setting without losing the deployment? Please find all the relevant logs at the following link: https://www.dropbox.com/sh/njl9aofhdw10ses/AADf2Ql4GKVIKcbgLivbmjC2a
Besides that, I believe this is a wrong behavior because HE should follow cluster properties (otherwise do not reach score 3400); do you believe is it worth opening a issue on bugzilla?
I would consider opening a BZ to track this. Adding some people who may have some insight on the issue.
The definition for the engine VM is getting extracted by ovirt-ha-agent from the OVF_STORE volume, not sure why the engine doesn't update it when you change cluster properties. I think it's work to fill a bug. Stefano, did you tried simply changing the number of cores for the engine VM from the engine to force a configuration update?
Thank you, Stefano.
_______________________________________________ Users mailing list Users@ovirt.org http://lists.ovirt.org/mailman/listinfo/users
-- Sandro Bonazzola Better technology. Faster innovation. Powered by community collaboration. See how it works at redhat.com
Hi Simone, I tried to change the number of cores and, after a reboot, the cpu model has changed to Nehalem. The HostedEngine vm is now migrating correctly. I will open a ticket anyway pointing to this specific behavior. Thank you very much for the support! -- Stefano Stagnaro Prisma Telecom Testing S.r.l. Via Petrocchi, 4 20127 Milano – Italy Tel. 02 26113507 int 339 e-mail: stefanos@prismatelecomtesting.com skype: stefano.stagnaro
participants (3)
-
Sandro Bonazzola -
Simone Tiraboschi -
Stefano Stagnaro