Re: RBD volume not made available to Xen virtual guest on openSUSE 15.2 (with libvirt 6.0.0)

On 10/22/20 6:48 PM, Marcel Juffermans wrote: > Hi there, > > Since upgrading to openSUSE 15.2 (which includes libvirt 6.0.0) the > virtual guests don't get their RBD disks made available to them. On > openSUSE 15.1 (which includes libvirt 5.1.0) that worked fine. The > XML is as follows: > > <domain type='xen' id='7'> >    <name>mytwotel-a</name> >    <uuid>a56daa5d-c095-49d5-ae1b-00b38353614e</uuid> >    <description>mytwotel-a</description> >    <memory unit='KiB'>1048576</memory> >    <currentMemory unit='KiB'>1048576</currentMemory> >    <vcpu placement='static'>1</vcpu> >    <cputune> >      <vcpupin vcpu='0' cpuset='8-23'/> >    </cputune> >    <os> >      <type arch='x86_64' machine='xenpv'>linux</type> > <kernel>/usr/lib/grub2/x86_64-xen/grub.xen</kernel> >    </os> >    <clock offset='utc'/> >    <on_poweroff>destroy</on_poweroff> >    <on_reboot>restart</on_reboot> >    <on_crash>restart</on_crash> >    <devices> >      <disk type='network' device='disk'> >        <driver name='qemu' type='raw' cache='writethrough'/> >        <source protocol='rbd' name='guests/mytwotel-a'> >        <auth username='libvirt'> >          <secret type='ceph' > uuid='3f88b59a-d85b-4b47-946d-a4c4cce3fec0'/> >        </auth> >        </source> >        <backingStore/> >        <target dev='xvda' bus='xen'/> >      </disk> >      <controller type='xenbus' index='0'/> >      <interface type='bridge'> >        <mac address='00:16:3e:a3:ba:9f'/> >        <source bridge='br0'/> >        <target dev='vif7.0'/> >      </interface> >      <console type='pty' tty='/dev/pts/0'> >        <source path='/dev/pts/0'/> >        <target type='xen' port='0'/> >      </console> >      <input type='mouse' bus='xen'/> >      <input type='keyboard' bus='xen'/> >      <memballoon model='xen'/> >    </devices> > </domain> > > The virtual guest starts, but then sits in the Grub 2 boot prompt > because the disk is not available. The qemu log shows: > > qemu-system-i386: failed to create 'qdisk' device '51712': failed to > create drive: Could not open > 'rbd:guests/mytwotel-a:id=libvirt:key=AQCAUpBbrcaiFxAA1sztXPbkdW1L54i99oUpyA==:auth_supported=cephx\;none': > No such file or directory > qemu-system-i386: failed to create 'qdisk' device '51712': failed to > create drive: Could not open > 'rbd:guests/mytwotel-a:id=libvirt:key=AQCAUpBbrcaiFxAA1sztXPbkdW1L54i99oUpyA==:auth_supported=cephx\;none': > No such file or directory > qemu-system-i386: failed to create 'qdisk' device '51712': failed to > create drive: Could not open > 'rbd:guests/mytwotel-a:id=libvirt:key=AQCAUpBbrcaiFxAA1sztXPbkdW1L54i99oUpyA==:auth_supported=cephx\;none': > No such file or directory > qemu-system-i386: failed to create 'qdisk' device '51712': failed to > create drive: Could not open > 'rbd:guests/mytwotel-a:id=libvirt:key=AQCAUpBbrcaiFxAA1sztXPbkdW1L54i99oUpyA==:auth_supported=cephx\;none': > No such file or directory > ... > > I tried to strace libvirtd. The results are as follows: > > On openSUSE 15.2 with libvirt 6.0.0 (not working), we see this: > > 1682  openat(AT_FDCWD, > "rbd:guests/mytwotel-a:id=libvirt:key=AQCAUpBbrcaiFxAA1sztXPbkdW1L54i99oUpyA==:auth_supported=cephx\\;none", > O_RDWR|O_CLOEXEC) = -1 ENOENT (No such file or directory) > 1682  rt_sigprocmask(SIG_BLOCK, NULL, [BUS USR1 ALRM IO], 8) = 0 > 1682  mmap(NULL, 1052672, PROT_READ|PROT_WRITE, > MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0x7f538aefd000 > 1682  mprotect(0x7f538aefd000, 4096, PROT_NONE <unfinished ...> > 1682  <... mprotect resumed>)           = 0 > 1682  rt_sigprocmask(SIG_SETMASK, [BUS USR1 ALRM IO], [BUS USR1 ALRM > IO], 8) = 0 > 1682  rt_sigprocmask(SIG_BLOCK, NULL, [BUS USR1 ALRM IO], 8) = 0 > 1682  mmap(NULL, 1052672, PROT_READ|PROT_WRITE, > MAP_PRIVATE|MAP_ANONYMOUS, -1, 0 <unfinished ...> > 1682  <... mmap resumed>)               = 0x7f538adfc000 > 1682  mprotect(0x7f538adfc000, 4096, PROT_NONE <unfinished ...> > 1682  <... mprotect resumed>)           = 0 > 1682  rt_sigprocmask(SIG_SETMASK, [BUS USR1 ALRM IO], <unfinished ...> > 1682  <... rt_sigprocmask resumed>[BUS USR1 ALRM IO], 8) = 0 > 1682  write(2, "qemu-system-i386: failed to crea"..., 232 <unfinished > ...> > ... > > On the other hand, on openSUSE 15.1 with libvirt 5.1.0 (working), we > see this: > > 16267 openat(AT_FDCWD, > "rbd:guests/mytwotel-a:id=libvirt:key=AQCAUpBbrcaiFxAA1sztXPbkdW1L54i99oUpyA==:auth_supported=cephx\\;none", > O_RDONLY|O_NONBLOCK|O_CLOEXEC) = -1 ENOENT (No such file or directory) > 16267 > stat("rbd:guests/mytwotel-a:id=libvirt:key=AQCAUpBbrcaiFxAA1sztXPbkdW1L54i99oUpyA==:auth_supported=cephx\\;none", > 0x7fff83e2e2b0) = -1 ENOENT (No such file or directory) > 16267 access("/usr/lib64/qemu/block-rbd.so", F_OK) = 0 > 16267 stat("/usr/lib64/qemu/block-rbd.so", {st_mode=S_IFREG|0644, > st_size=27448, ...}) = 0 > 16267 openat(AT_FDCWD, "/usr/lib64/qemu/block-rbd.so", > O_RDONLY|O_CLOEXEC) = 60 > 16267 read(60, "\177ELF\2\1\1\0\0\0\0\0\0\0\0\0\3\0>\0\1\0\0\0 > &\0\0\0\0\0\0"..., 832) = 832 > 16267 fstat(60, {st_mode=S_IFREG|0644, st_size=27448, ...}) = 0 > 16267 mmap(NULL, 2122672, PROT_READ|PROT_EXEC, > MAP_PRIVATE|MAP_DENYWRITE, 60, 0) = 0x7f8e6030f000 > 16267 mprotect(0x7f8e60315000, 2093056, PROT_NONE) = 0 > 16267 mmap(0x7f8e60514000, 8192, PROT_READ|PROT_WRITE, > MAP_PRIVATE|MAP_FIXED|MAP_DENYWRITE, 60, 0x5000) = 0x7f8e60514000 > 16267 close(60)                         = 0 > ... > > Note that the latter opens "/usr/lib64/qemu/block-rbd.so". That > library *does* exist on openSUSE 15.2 but it doesn't seem to be used. It smells like a bug in the Leap 15.2 qemu package. Or perhaps in xen's libxl library, which controls the qemu processes used by VMs. I'd suggest entering a bug at bugzilla.opensuse.org. > I've tried to update libvirt to a newer version using the Open Build > Service repos, but then ran into so many conflicting versions that I > gave up. > > At this point I'm stuck. Does anyone have an idea I can try? We might be able to eliminate libxl by comparing the interaction with qemu between the working and non-working setups. E.g. enable debug in /etc/libvirt/libvirtd.conf, restart libvirtd, start the offending VM, then look in /var/log/libvirt/libxl/libxl-driver.log for the qemu command line created by libxl and qmp commands sent to qemu. If libxl prepares qemu similarly between the two, then the issue is likely in the qemu package. Regards, Jim