Hi,
I have been experimenting with VM snapshots on XenServer and have noticed a
problem that I hope someone might be able to shed some light on.
In a normal flow of taking a VM snapshot, reverting to it, then deleting
the VM snapshot, I have observed the following (which looks just fine):
*SR:*
uuid ( RO) : 2a061111-a8c8-11db-4a8b-f8d519f9ac3e
name-label ( RW): Test
name-description ( RW): iSCSI SR [10.10.8.108
(iqn.2010-01.com.solidfire:3y8w.test.15; LUN 0:
337938770000000ff47acc0100000000: 93.1 GB (SolidFir))]
host ( RO): XenServer-6.1-Tut-2
type ( RO): lvmoiscsi
content-type ( RO):
- *Before VM snap:*
*Active:*
uuid ( RO) : b4587018-9679-4fe7-ba72-5523cb988cec
name-label ( RW): i-2-21-VM-DATA
name-description ( RW):
sr-uuid ( RO): 2a061111-a8c8-11db-4a8b-f8d519f9ac3e
virtual-size ( RO): 16106127360
sharable ( RO): false
read-only ( RO): false
- *After VM snap:*
*Base copy (contains the data of the previously active VDI):*
uuid ( RO) : d167952d-deb4-4942-9ea8-c8b3777d885e
name-label ( RW): base copy
name-description ( RW):
sr-uuid ( RO): 2a061111-a8c8-11db-4a8b-f8d519f9ac3e
virtual-size ( RO): 16106127360
sharable ( RO): false
read-only ( RO): true
*Snapshot:*
uuid ( RO) : 613dc799-cf69-445a-a2fe-611653e0b0c9
name-label ( RW): i-2-21-VM-DATA
name-description ( RW):
sr-uuid ( RO): 2a061111-a8c8-11db-4a8b-f8d519f9ac3e
virtual-size ( RO): 16106127360
sharable ( RO): false
read-only ( RO): false
*Active (has the same UUID as the previously active VDI):*
uuid ( RO) : b4587018-9679-4fe7-ba72-5523cb988cec
name-label ( RW): i-2-21-VM-DATA
name-description ( RW):
sr-uuid ( RO): 2a061111-a8c8-11db-4a8b-f8d519f9ac3e
virtual-size ( RO): 16106127360
sharable ( RO): false
read-only ( RO): false
- *After revert to VM snap:*
*Base copy:*
uuid ( RO) : d167952d-deb4-4942-9ea8-c8b3777d885e
name-label ( RW): base copy
name-description ( RW):
sr-uuid ( RO): 2a061111-a8c8-11db-4a8b-f8d519f9ac3e
virtual-size ( RO): 16106127360
sharable ( RO): false
read-only ( RO): true
*Snapshot (this VDI is un-touched):*
uuid ( RO) : 613dc799-cf69-445a-a2fe-611653e0b0c9
name-label ( RW): i-2-21-VM-DATA
name-description ( RW):
sr-uuid ( RO): 2a061111-a8c8-11db-4a8b-f8d519f9ac3e
virtual-size ( RO): 16106127360
sharable ( RO): false
read-only ( RO): false
*Active (this is a new VDI - the old active VDI was deleted):*
uuid ( RO) : b21284fa-347a-459a-a8bf-0fcd7717a134
name-label ( RW): i-2-21-VM-DATA
name-description ( RW):
sr-uuid ( RO): 2a061111-a8c8-11db-4a8b-f8d519f9ac3e
virtual-size ( RO): 16106127360
sharable ( RO): false
read-only ( RO): false
- *After deleting VM snap:*
*Active (the snapshot is gone as is the base copy...the base copy was
rolled up into this VDI):*
uuid ( RO) : b21284fa-347a-459a-a8bf-0fcd7717a134
name-label ( RW): i-2-21-VM-DATA
name-description ( RW):
sr-uuid ( RO): 2a061111-a8c8-11db-4a8b-f8d519f9ac3e
virtual-size ( RO): 16106127360
sharable ( RO): false
read-only ( RO): false
Now, in my case, where I create an SR on the fly (in response to attaching
a CloudStack volume to a VM on XenServer for the first time) to house a
single VDI (which has guaranteed IOPS), I see the following erroneous
behavior when it comes to hypervisor snapshots:
*SR:*
uuid ( RO) : 70e06f08-2c9d-f9cf-4e64-2f064c11325a
name-label ( RW): /iqn.2010-01.com.solidfire:3y8w.test.19/0
name-description ( RW): /iqn.2010-01.com.solidfire:3y8w.test.19/0
host ( RO): XenServer-6.1-Tut-2
type ( RO): lvmoiscsi
content-type ( RO): user
- *Before VM snap:*
*Active:*
uuid ( RO) : 067572a8-fa4d-45b5-9365-2d7790a4b202
name-label ( RW): i-2-21-VM-DATA
name-description ( RW):
sr-uuid ( RO): 70e06f08-2c9d-f9cf-4e64-2f064c11325a
virtual-size ( RO): 10737418240
sharable ( RO): false
read-only ( RO): false
- *After VM snap (this appears just fine):*
*Base copy:*
uuid ( RO) : 71d39b5b-6c90-4aa9-adbf-71b226652081
name-label ( RW): base copy
name-description ( RW):
sr-uuid ( RO): 70e06f08-2c9d-f9cf-4e64-2f064c11325a
virtual-size ( RO): 10737418240
sharable ( RO): false
read-only ( RO): true
*Snapshot:*
uuid ( RO) : afc66ec7-5493-4772-9318-6f72c9d971f8
name-label ( RW): i-2-21-VM-DATA
name-description ( RW):
sr-uuid ( RO): 70e06f08-2c9d-f9cf-4e64-2f064c11325a
virtual-size ( RO): 10737418240
sharable ( RO): false
read-only ( RO): false
*Active:*
uuid ( RO) : 067572a8-fa4d-45b5-9365-2d7790a4b202
name-label ( RW): i-2-21-VM-DATA
name-description ( RW):
sr-uuid ( RO): 70e06f08-2c9d-f9cf-4e64-2f064c11325a
virtual-size ( RO): 10737418240
sharable ( RO): false
read-only ( RO): false
- *After a failed revert to VM snap:*
uuid ( RO) : afc66ec7-5493-4772-9318-6f72c9d971f8
name-label ( RW): i-2-21-VM-DATA
name-description ( RW):
sr-uuid ( RO): 70e06f08-2c9d-f9cf-4e64-2f064c11325a
virtual-size ( RO): 10737418240
sharable ( RO): false
read-only ( RO): false
Somehow the base copy and active VDI have both been deleted and the
snapshot VDI is the only remaining VDI.
I would have expected the active VDI to be deleted and a new VDI (which is
initially empty) to take its place as the active VDI. The snapshot should
not be touched and the base copy should not be deleted.
Does anyone have any insight as to why this may be happening?
Thanks!
--
*Mike Tutkowski*
*Senior CloudStack Developer, SolidFire Inc.*
e: mike.tutkow...@solidfire.com
o: 303.746.7302
Advancing the way the world uses the
cloud<http://solidfire.com/solution/overview/?video=play>
*™*
