Symptoms
All VEs provisioning and recreate operations are failing on particular Virtuozzo node. The following error message could be found in PACI-im.log right after VE creation:
ERROR GenericVePcsTask [Shared executor thread #2 @4] - No storage id for PCS drive
On Vm2vf side, the following notable structure is displayed at the end of creation process:
Invoked callback [done_with_capacity_and_id(0, 37435960672256, 1466097127270, null, 3e5cdcd0-f7ef-4845-9b29-f4a9012710c7)]
Note 'null' value in the line above. There should be location of CT or VM private space. When this callback is received, VE considered as corrupted and gets immediately deleted.
Cause
VM home directory is configured to the non-default location:
[root@pcs ~]# prlsrvctl info | grep 'VM home'
VM home: /pcs
Resolution
According to OA documentation, when Virtuozzo node operates in cluster environment, default directory should be used for VM home
.
Please note that for VZ6 and VZ7 directories should be different:
VZ6: /var/parallels
VZ7: /vz/vmprivate
You can correct this setting the following way.:
# prlsrvctl user set --def-vm-home /vz/vmprivate
# prlsrvctl user list
Then edit file /etc/vz/vz.conf and change value for VE_PRIVATE
from:
VE_PRIVATE="/pcs/$VEID"
to
VE_PRIVATE=/vz/private/$VEID
Perform these operations on all nodes where you find incorrect values for VM home
. Please note that OACI IM restart and PACI-vm2vf might be needed as described in KB article.