On Mon, Jun 8, 2020 at 6:23 PM Gianluca Cecchi <gianluca.cec...@gmail.com> wrote: > > Just to give a report of final success updating from 4.3.7 to 4.4 one of my > home labs, composed by a single host. > > hw: Intel NUC6i5SY with 32G of ram and 2 SSD disks (250Gb and 500Gb) > > source sw: oVirt 4.3.7 single host with CentOS 7 OS and storage provided > through the host itself via NFS (not officially supported, but working, apart > when shutdown needed) > Two main VMs to migrate to the new environment: Fedora 30 and Slackware > Current (just to not forget the first love... ;-)
:-) > > Exported the VMs to an export storage domain offered through an external USB > disk Any particular reason to do this instead of upgrade via engine backup/restore? Each has its own Pros and Cons, of course. > > dest sw: oVirt node ng 4.4 configured with HCI single host wizard. Installed > on the 250Gb disk. I pre-clean the disks (dd of the first 100Mb of the disks) > before install, because during the beta/rc phase I noticed the installer was > not so smart to cleanup pre-existing configurations. > I had problems in the first run, but with engine cleanup and redeploying it > went ok. > See here for more details: > https://lists.ovirt.org/archives/list/users@ovirt.org/thread/6QODLB6J5Z74YCVF6C3TLQPF4KK7RKB5/ The issue there (admin password with '%') should be fixed in 4.4.1, soon: https://bugzilla.redhat.com/show_bug.cgi?id=1839533 > > I configured the gluster domains in the wizard on the whole second disk. > I then imported the 2 VMs without problems. Before starting them up I changed > their inherited "BIOS Type" from Legacy to "Default Cluster" and they both > started without any problem. > While in 4.4 I was able to update Fedora VM from 30 to 31 and then 32 and > also refresh the slackware-current one that was about a month behind from > latest current. > With "Default Cluster" BIOS type the VMs start with the following options: > " -machine pc-q35-rhel8.1.0,accel=kvm,usb=off,dump-guest-core=off -cpu > Broadwell-noTSX" > > I was then able to download a CentOS 8 cloud image from the predefined > ovirt-image-repository storage domain and convert it to template. > Created a VM from this template and cloud-init was able to inject the ssh > public key and set timezone. > Changed cluster type of ths VM to Default as above with a warning from oVirt > but without any problem inside the VM and also changed the disk type from > virtio to virtio-scsi without problems. > Now I enabled incremental backup at engine level and also at disk level of > this CentOS 8 VM so that I can make some tests in this regard. > > Also, I was able to successfully test the ovirt-ansible-shutdown-env ansible > role to make a clean overall shutdown of the environment, one of the things > that in my previous unsupported setup was a little cumbersome. > > Right after install I noticed that in CentOS 8 ovirt-node-ng setup, > intel_pstate was the default scaling driver setup, but my cpu was almost > always crying with core at 2.6Ghz (and temp around 90 degrees) even if > attempted to setup conservative profiles and without nothing running, apart > engine VM. > The NUC is under my working day table and I don't need always performance > from it... > So I modified (notice also the rhgb and quite omissions...) these files and > reboot: > > - /etc/default/grub > GRUB_CMDLINE_LINUX='crashkernel=auto resume=/dev/mapper/onn-swap > rd.lvm.lv=onn/ovirt-node-ng-4.4.0-0.20200521.0+1 rd.lvm.lv=onn/swap > intel_pstate=disable' > > - /boot/grub2/grub.cfg > set default_kernelopts="root=UUID=85212719-8feb-43aa-9819-2820d4672795 ro > crashkernel=auto ipv6.disable=1 intel_pstate=disable " > > - > /boot/loader/entries/ovirt-node-ng-4.4.0-0.20200521.0+1-4.18.0-147.8.1.el8_1.x86_64.conf > options intel_pstate=disable boot=UUID=b717ab4f-ca71-469a-8836-ff92cebc7650 > crashkernel=auto rd.lvm.lv=onn/swap > root=/dev/onn/ovirt-node-ng-4.4.0-0.20200521.0+1 resume=/dev/mapper/onn-swap > rootflags=discard rd.lvm.lv=onn/ovirt-node-ng-4.4.0-0.20200521.0+1 > img.bootid=ovirt-node-ng-4.4.0-0.20200521.0+1 null > > - /boot/efi/EFI/centos/grub.cfg > set default_kernelopts="root=/dev/mapper/onn-root ro crashkernel=auto > resume=/dev/mapper/onn-swap rd.lvm.lv=onn/root rd.lvm.lv=onn/swap > intel_pstate=disable " > > And now with the default/old scaling driver all is quite silent and still > working ok for my needs, with web admin quite usable. I updated my two VMs as > described above after cpompleting these changes. > Right now for example I have hosted engine running + slackware one + CentOS 8 > one with > > [root@ovirt01 ~]# cat /proc/cpuinfo | grep Hz > model name : Intel(R) Core(TM) i5-6260U CPU @ 1.80GHz > cpu MHz : 648.431 > model name : Intel(R) Core(TM) i5-6260U CPU @ 1.80GHz > cpu MHz : 628.307 > model name : Intel(R) Core(TM) i5-6260U CPU @ 1.80GHz > cpu MHz : 648.859 > model name : Intel(R) Core(TM) i5-6260U CPU @ 1.80GHz > cpu MHz : 663.792 > [root@ovirt01 ~]# > > and > [root@ovirt01 ~# cat /sys/devices/system/cpu/cpu*/cpufreq/scaling_governor > powersave > powersave > powersave > powersave > [root@ovirt01 g.cecchi]# > > Thanks to all that helped throughout ... Thanks for the report! Best regards, -- Didi _______________________________________________ Users mailing list -- users@ovirt.org To unsubscribe send an email to users-le...@ovirt.org Privacy Statement: https://www.ovirt.org/privacy-policy.html oVirt Code of Conduct: https://www.ovirt.org/community/about/community-guidelines/ List Archives: https://lists.ovirt.org/archives/list/users@ovirt.org/message/TSBSQUFNXKOG2S5BQPICRZB6NH7GTISO/
