postcopy migration isn't the default but can in some cases be very useful
With Ubuntu 21.04 Hirsute kernel being v5.11 postcopy migration will fail like:
virsh migrate --unsafe --live --postcopy --postcopy-after-precopy kvmguest-focal-postcopy qemu+ssh://10.85.93.248/system
error: internal error: unable to execute QEMU command 'migrate-set-capabilities': Postcopy is not supported
I was using a setup using FCP-disks -> Multipath -> LVM not being mounted anymore after an upgrade from 18.04 to 20.04.
I was seeing these errors at boot - I thought that is ok to sort out duplicates:
May 28 09:00:43 s1lp05 lvm[746]: WARNING: Not using device /dev/sdd1 for PV q1KTMM-fkpM-Ewvm-T4qd-WgO8-hV79-qXpUpb.
May 28 09:00:43 s1lp05 lvm[746]: WARNING: Not using device /dev/sdi1 for PV q1KTMM-fkpM-Ewvm-T4qd-WgO8-hV79-qXpUpb.
May 28 09:00:43 s1lp05 lvm[746]: WARNING: Not using device /dev/sdn1 for PV q1KTMM-fkpM-Ewvm-T4qd-WgO8-hV79-qXpUpb.
May 28 09:00:43 s1lp05 lvm[746]: WARNING: PV q1KTMM-fkpM-Ewvm-T4qd-WgO8-hV79-qXpUpb prefers device /dev/sds1 because device was seen first.
May 28 09:00:43 s1lp05 lvm[746]: WARNING: PV q1KTMM-fkpM-Ewvm-T4qd-WgO8-hV79-qXpUpb prefers device /dev/sds1 because device was seen first.
May 28 09:00:43 s1lp05 lvm[746]: WARNING: PV q1KTMM-fkpM-Ewvm-T4qd-WgO8-hV79-qXpUpb prefers device /dev/sds1 because device was seen first.
But then later auto-activation fails on duplicates
May 28 09:00:56 s1lp05 systemd[1]: Starting LVM event activation on device 253:8...
May 28 09:00:56 s1lp05 lvm[1882]: pvscan[1882] PV /dev/mapper/mpathd-part1 is duplicate for PVID q1KTMMfkpMEwvmT4qdWgO8hV79qXpUpb on 253:8 and 8:49.
May 28 09:00:57 s1lp05 systemd[1]: lvm2-pvscan@253:8.service: Main process exited, code=exited, status=5/NOTINSTALLED
May 28 09:00:57 s1lp05 systemd[1]: lvm2-pvscan@253:8.service: Failed with result 'exit-code'.
May 28 09:00:57 s1lp05 systemd[1]: Failed to start LVM event activation on device 253:8.
And finally giving up:
May 28 09:02:12 s1lp05 systemd[1]: dev-vgdisks-lv_tmp.device: Job dev-vgdisks-lv_tmp.device/start timed out.
lvdisplay reported the devices then as LV Status NOT available
It seems LVM now scans more (or the kernel presents more) devices. I didn't have these issues on 18.04.
In a discussion spawned from the issue to connect to libvirt through ssh we discussed about using libvirt with the usual default of qemu://session right after installing libvirt* packages as there are some hurdles to pass - those hurdles do not exist as soon as you re-login btw.
What will happen is like:
$ apt install libvirt-daemon-system
$ virsh list
error: failed to connect to the hypervisor
error: Failed to connect socket to '/var/run/libvirt/libvirt-sock':
Permission denied
But actually slightly more sinister is when it works but you later miss your guests. Which is due to the default right after install being from the binaries which is qemu:///session (in newer versions where this was changed).
$ apt install libvirt-daemon-system $ virsh uri qemu:///session
Why is my guest failing to start / failing to attach a device with this apparmor profile error.
Start:
error: Failed to start domain c
error: internal error: cannot load AppArmor profile 'libvirt-8763795d-f056-4373-9d57-3d3bad391e5a'
attach-device:
error: Failed to attach device from disk.xml
error: internal error: cannot update AppArmor profile 'libvirt-8763795d-f056-4373-9d57-3d3bad391e5a'
Where and what is this libvirt-8763795d-f056-4373-9d57-3d3bad391e5a thing?
I mostly found bcache to suite me better, but due to a discussion today I came to test dm-cache. I didn't find a great entry doc how to do so, so I thought I could as well document and share here - and by that make "searchable".
So how to actually set dm-cache up on Ubuntu?