[pve-devel] Possible bug in pve-manager and/or qemu-kvm

[Michael Rasmussen]

Hi all,

I have noticed a possible bug in pve-manager and/or qemu-kvm.

Scenario.
You have a VM running and now you want to should it down. To do this
you can choose either to use the shutdown function from the gui or the
console or use the client native shutdown method. This all works as
expected except for one, IMHO, important situation:

Using gui shutdown or console shutdown will generate this in syslog

Jun 15 21:37:19 pve pvedaemon[4842]: <root at pam> end task UPID:pve:00001FEE:0008B
C24:5942E1EB:qmstart:110:root at pam: OK
Jun 15 21:37:43 pve pvedaemon[5746]: <root at pam> starting task UPID:pve:0000203E:
0008C709:5942E207:qmshutdown:110:root at pam:
Jun 15 21:37:43 pve pvedaemon[8254]: shutdown VM 110: UPID:pve:0000203E:0008C709
:5942E207:qmshutdown:110:root at pam:
Jun 15 21:37:45 pve kernel: [ 5754.731169]  sdb: sdb1 sdb2 < sdb5 >
Jun 15 21:37:45 pve kernel: [ 5755.119112] vmbr0: port 2(tap110i0) entered disabled state
Jun 15 21:37:45 pve pvedaemon[5746]: <root at pam> end task UPID:pve:00001FFB:0008BCA6:5942E1ED:vncproxy:110:root at pam: OK
Jun 15 21:37:45 pve kernel: [ 5755.333909] sd 33:0:0:0: [sdb] Synchronizing SCSI cache
Jun 15 21:37:45 pve pvedaemon[5746]: <root at pam> end task UPID:pve:0000203E:0008C709:5942E207:qmshutdown:110:root at pam: OK
Jun 15 21:37:45 pve iscsid: Connection1:0 to [target: iqn.2005-10.org.freenas.ctl:vm-110, portal: 10.0.1.32,3260] through [iface: default] is shutdown.

As above shows the deactivate_volume is called for every disk
belonging to the VM.

Using native client to shutdown will generate this is syslog

Jun 15 21:37:46 pve qm[8308]: VM 110 qmp command failed - VM 110 not running

As above shows the deactivate_volume is not called for any disk
belonging to the VM so every disk is left activated.

I should mention that for CT's there is no differens whether gui
shutdown or native shutdown is used.

I wonder if the process login this 'pve qm[8308]: VM 110 qmp command
failed - VM 110 not running' should not call deactivate_volume for every disk belonging to the VM?

-- 
Hilsen/Regards
Michael Rasmussen

Get my public GnuPG keys:
michael <at> rasmussen <dot> cc
http://pgp.mit.edu:11371/pks/lookup?op=get&search=0xD3C9A00E
mir <at> datanom <dot> net
http://pgp.mit.edu:11371/pks/lookup?op=get&search=0xE501F51C
mir <at> miras <dot> org
http://pgp.mit.edu:11371/pks/lookup?op=get&search=0xE3E80917
--------------------------------------------------------------
/usr/games/fortune -es says:
Raising pet electric eels is gaining a lot of current popularity.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 488 bytes
Desc: OpenPGP digital signature
URL: <http://lists.proxmox.com/pipermail/pve-devel/attachments/20170615/77f20c58/attachment.sig>