On 01/17/2017 02:13 PM, Marc Hartmayer wrote:
Update: It's a SELinux labeling problem and seems to be introduced by the QEMU namespace patches.
I wouldn't guess from the error message that qemu is getting EPERM.
Anyway, the SELinux issue is fixed in -rc2:
commit 93a062c3b293685024d60e841a37e93e303f4943 Author: Michal Privoznik <mprivozn@redhat.com> AuthorDate: Fri Jan 13 10:03:23 2017 +0100 Commit: Michal Privoznik <mprivozn@redhat.com> CommitDate: Fri Jan 13 14:45:52 2017 +0100
qemu: Copy SELinux labels for namespace too
When creating new /dev/* for qemu, we do chown() and copy ACLs to create the exact copy from the original /dev. I though that copying SELinux labels is not necessary as SELinux will chose the sane defaults. Surprisingly, it does not leaving namespace with the following labels:
crw-rw-rw-. root root system_u:object_r:tmpfs_t:s0 random crw-------. root root system_u:object_r:tmpfs_t:s0 rtc0 drwxrwxrwt. root root system_u:object_r:tmpfs_t:s0 shm crw-rw-rw-. root root system_u:object_r:tmpfs_t:s0 urandom
As a result, domain is unable to start:
error: internal error: process exited while connecting to monitor: Error in GnuTLS initialization: Failed to acquire random data. qemu-kvm: cannot initialize crypto: Unable to initialize GNUTLS library: Failed to acquire random data.
The solution is to copy the SELinux labels as well.
Reported-by: Andrea Bolognani <abologna@redhat.com> Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
On Tue, Jan 17, 2017 at 11:18 AM +0100, Marc Hartmayer <mhartmay@linux.vnet.ibm.com> wrote:
Hey,
I have tried to live hot plug a disk backed on a qcow2 disk (see XML snippet below) on a s390 system and I've got the following error message:
<error_message> internal error: unable to execute QEMU command 'device_add': Property 'scsi-hd.drive' can't find value 'drive-scsi0-0-0-0' </error_message>
<xml_snippet> <disk type="file"> <driver name="qemu" type="qcow2"/> <source file="/tmp/virtd-test_e3hnhh5/disk1.qcow2" />
My namespace patches should not clash with this as this isn't a device from /dev/*. In the namespace, it's just /dev that is different to the parent namespace. So anything else (e.g. under /tmp) is "shared" with the parent namespace (it is the same mount in fact).
<target bus="scsi" dev="sda" /> </disk> </xml_snippet>
With v2.5.0 everything has worked. I'll take a closer look to it today.
You can try and see if this is a namespace caused issue. Just disable the namespaces and retry. If it succeeds with namespaces disabled, the bug indeed is in my namespaces patches.
btw: to disable namespaces set: namespaces=[] in /etc/libvirt/qemu.conf
Michal Somewhat related: I built 3.0.0-rc2 on an x86 Ubuntu box and could not get it to run in conjunction with apparmor. By this I mean I could not start a simple KVM guest. I could get past the audit failures by adding an unconditional mount
On 17.01.2017 14:21, Michal Privoznik wrote: permission in /etc/apparmor.d/usr.sbin.libvirtd but ended up with a generic failure message doing a virsh start. Disabling namespaces helps (obviously). It would be nice if someone apparmor-savvy could have a look. Thanks!
-- libvir-list mailing list libvir-list@redhat.com https://www.redhat.com/mailman/listinfo/libvir-list
-- Mit freundlichen Grüßen/Kind Regards Viktor Mihajlovski IBM Deutschland Research & Development GmbH Vorsitzender des Aufsichtsrats: Martina Köderitz Geschäftsführung: Dirk Wittkopp Sitz der Gesellschaft: Böblingen Registergericht: Amtsgericht Stuttgart, HRB 243294