Cannot do online Migration for CentOS 6 + Plesk template

[Machiasiaweb]

Hello,

Did you have experience when doing online migration under PVC 4.7 with CentOS6 + Plesk 10 with error?

=========
# vzmigrate --online --keep-dst n4 CTID
Connection to destination node (n4) is successfully established
Moving/copying CT#CTID -> CT#CTID, [], [] ...
Checking external bind mounts
Check cluster ID
Checking keep dir for private area copy
Checking SLM-only mode
Checking CPT image version for online migration
Checking capabilities for online migration
Checking license restrictions
Checking technologies
Checking templates for CT
copy ez template area directories
Checking caches
Checking IP addresses on destination node
Tracker started
Syncing private area '/vz/private/CTID'
done
OfflineManagement CT#CTID ...
done
Suspending CT#CTID ...
done
Dumping CT#CTID ...
done
Syncing dump file...
done
Syncing tracked files from '/vz/private/CTID/fs'
done
/usr/sbin/vzctl exited with code 17
can not undump CT#CTID : /usr/sbin/vzctl exited with code 17
This CT can't be migrated online at the moment. Try offline migration.
Dumpfile /vz/dump/dumpfile.OsLoj0.saved saved
Can't move/copy CT#CTID -> CT#CTID, [], [] : /usr/sbin/vzctl exited with code 17
Resuming CT#CTID ...
done
OfflineManagement CT#CTID ...
done
=============

Did anyone know how to solve it?

Thx!

[KristianM]

Hi,

are source and destination node both running VZ 4.7?
Same kernel?

Does the CT have any special settings such as NFS support on? etc?
Returncode 17 is a "Checkpoint restore error".
I would guess the destination node doesn't meet all requirements
for the CT to restore.

Regards
Kristian

[Machiasiaweb]

Hello,

Both node are running VZ 4.6 with same kernel. NFS did not enabled. I was found some fore details when using -v to do vzmigrate

....
..
–dumpfile /vz/dump/dumpfile.3BPuQp –context 180f
1331712744: 1331712744: vzctl : Can't undump: Not a directory
1331712744: 1331712744: vzctl : Error: Failed to restore mount point: dev '/vz/template:/vz/private/VEID/fs', type
'vzfs', path '/var/named/chroot/etc/named.rfc1912.zones'
..

After that I stopped the named services and then do vzmigrate again. It work!!

But I am sure did not have such problem when CentOS5 + Plesk 10 before.

Did any idea? Thanks!

[KristianM]

Hi,

I'd guess it has something to do with the chroot of bind with CentOS 6.
Not sure but I think with CentOS 5 there was no chroot?

Try filling a bug with Parallels
Good luck while you're at it!

Regards
Kristian

[figjam]

Hi,

I'd guess it has something to do with the chroot of bind with CentOS 6.
Not sure but I think with CentOS 5 there was no chroot?

Interesting that this was one of the fixes in the last PVC 4.0 & 4.6 kernel update (CU-2.6.18-028stab099.3).

- An online migration of a Container with bindmounted files inside could fail. In particular, bindmounted files are used by modern versions of the "named" service. (PCLIN-30785)

Maybe it hasn't made it into PVC 4.7 yet?

Cheers,


Andrew

[figjam]

CU-2.6.32-042stab053.5 Parallels Virtuozzo Containers for Linux 4.7 kernel update

* An online migration of a Container with bindmounted files inside could fail.
In particular, bindmounted files are used by modern versions of the "named"
service. (PCLIN-30771)


Cheers,


Andrew

[Machiasiaweb]

Hi,

Thanks, but issue another new problem when I migrate from PVC 4.7 to PVC 4.6 node.

# vzmigrate --online --keep-dst another_node CTID
Connection to destination node (another_node) is successfully established
Moving/copying CT#CTID -> CT#CTID, [], [] ...
Checking external bind mounts
Check cluster ID
Checking keep dir for private area copy
Checking license restrictions
Check of requires kernel modules
Checking CPT image version for online migration
/usr/libexec/vztestver exited with code 1
Can't move/copy CT#CTID -> CT#CTID, [], [] : /usr/libexec/vztestver exited with code 1

Finally require to do migration using offline.

Reverse migrate from PVC 4.6 to PVC 4.7 still can using online migration.

Please see any experience like you have.

Thanks!

[figjam]

Can you add "--verbose" to the parameters you are using and post the failure output here?

Cheers,


Andrew

[Machiasiaweb]

Hello,

That is the output
==============================

1334802701: ssh -T -q -c arcfour -o StrictHostKeyChecking=no -o CheckHostIP=no -o UserKnownHostsFile=/dev/null -o PreferredAuthentications=publickey,password,keyboa rd-interactive nodeA true
1334802701: ssh -T -q -c arcfour -o StrictHostKeyChecking=no -o CheckHostIP=no -o UserKnownHostsFile=/dev/null -o PreferredAuthentications=publickey,password,keyboa rd-interactive nodeA /usr/sbin/vzmdest 470 -v --online '7772:7772'
1334802702: remote_version 460
1334802702: Connection to destination node (nodeA) is successfully established
1334802702: Moving/copying CT#CTID -> CT#CTID, [], [] ...
1334802702: begin stage : preMigrateStage
1334802702: Send command: init %d %d
1334802702: 1334802702: Command : init
1334802702: 1334802702: begin stage : initVEMigration
1334802702: 1334802702: Start of CT CTID migration (private /vz/private/CTID, root /vz/root/CTID)
1334802702: 1334802702: add 'on failure' remove cleaner : /vz/root/CTID
1334802702: 1334802702: end stage : initVEMigration
1334802702: Checking external bind mounts
1334802702: Check cluster ID
1334802702: checkoptions 0
1334802702: Send command: checkoptions 0
1334802702: 1334802702: Command : checkoptions
1334802702: establish ssh swap channel: /usr/libexec/vziterind CTID
1334802702: ssh -T -q -c arcfour -o StrictHostKeyChecking=no -o CheckHostIP=no -o UserKnownHostsFile=/dev/null -o PreferredAuthentications=publickey,password,keyboa rd-interactive vz-c01n4 /usr/libexec/vziterind CTID
1334802702: Swap channel established
1334802702: Checking keep dir for private area copy
1334802702: Send command: checkkeepdir
1334802702: 1334802702: Command : checkkeepdir
1334802702: Checking license restrictions
1334802702: Send command: checklicense
1334802702: 1334802702: Command : checklicense
1334802702: 1334802702: checkLicense: 151 (122)
1334802702: Check of requires kernel modules
1334802702: Checking CPT image version for online migration
1334802702: Send command: cpt_ver %s
1334802702: 1334802702: Command : cpt_ver
1334802702: 1334802702: /usr/libexec/vztestver 2048
1334802702: 1334802702: vztestver : image cannot be restored
1334802702: 1334802702: /usr/libexec/vztestver exited with code 1
1334802702: 1334802702: CPT image cannot be restored
1334802702: 1334802702:
This CT can't be migrated online at the moment. Try offline migration.
1334802702: 1334802702: error [-2] : /usr/libexec/vztestver exited with code 1
1334802702: /usr/libexec/vztestver exited with code 1
1334802702: Can't move/copy CT#CTID -> CT#CTID, [], [] : /usr/libexec/vztestver exited with code 1
1334802702: cleaning : close ssh channel
1334802702: close channel

==========================
Thx!

[figjam]

Hi,

Can you try the following?

• On the source node:
  
  • Uncomment SCAN_PRIVATE="yes" from /etc/vztt/vztt.conf
  • Execute: vzpkg sync_vzpackages <CTID>
  
  
• Uncomment SCAN_PRIVATE="yes" from /etc/vztt/vztt.conf on the destination node
• Migrate the container

If it works, comment those lines out again in /etc/vztt/vztt.conf on both nodes. I haven't had a failure with this method since KristianM put me onto it in another thread.

Cheers,


Andrew

[Machiasiaweb]

Thanks for your suggestion, but no luck for me.

[paulieG]

Hi,

I think you're going to struggle to get this working as its specifically mentioned as not working : http://www.parallels.com/fileadmin/p...inuxREADME.pdf (In Known Issues).

It specifically states online migration from 4.7 to 4.6 is unsupported so by implication offline migration should work.

Paul.

[Machiasiaweb]

Hello,

Update this ticket again as just setup another node with PVC 4.7. No luck and find some strange. Simple evnirnment summary:

Server A: PVC 4.7, CentOS 5.8, kernel 2.6.32-042stab053.5

Server B: PVC 4.7, RHEL 6.2, kernel 2.6.32-042stab055.10

Migration from Server B with to Server A online is fail

but

Migration from Server A with online to Server B is FINE

Those error are same as the pervious. Did you have any idea?

[figjam]

Server A: PVC 4.7, CentOS 5.8, kernel 2.6.32-042stab053.5

Server B: PVC 4.7, RHEL 6.2, kernel 2.6.32-042stab055.10

Migration from Server B with to Server A online is fail

but

Migration from Server A with online to Server B is FINE

That is because Server B has a newer kernel than Server A. It is not possible to do an online migration to an older kernel version.

Cheers,


Andrew