[Users] Container Migration - Ploop

Kir Kolyshkin kir at openvz.org
Mon Mar 4 19:42:47 EST 2013


On 03/04/2013 04:00 PM, Axton wrote:
> On Sun, Mar 3, 2013 at 6:21 PM, Axton <axton.grams at gmail.com> wrote:
>> I am attempting to migrate a container from one host to another.  I am
>> running into issues that cause the migration to fail.
>>
>> Environment information:
>> - Build:Debian 6; updates, install RHEL kernel
>> - Kernel:
>> 042stab072.10
>>
>> The only part that is failing is the live migration.  If I stop the
>> container, I can migrate without any issue.
>>
>> I will try to look more into this, but wanted to see if it is a known issue.
>>
>> Axton Grams
> The two machines were built identically and are patched at the same
> level, have the same hardware configuration, same drive and file
> system configuration.  I did some things with the file systems to put
> things into an optimal arrangement.  Here are the details of the file
> system configuration:
>
> Looking further at this, it appears that the vzctl 4.1.2 snapshot
> located at http://download.openvz.org/utils/vzctl/4.1.2/vzctl-4.1.2-1.x86_64.rpm
> does not support the chkpnt option.  I tried the following command,
> and it returns the same error that the live migration returns:
>
> root at cluster-01:/var/log# vzctl chkpnt 5000 --dumpfile /tmp/5000.dump
> Setting up checkpoint...
>          suspend...
>          dump...
> Can not dump container: Invalid argument
> Checkpointing failed
>
> /var/log/kern.log contains the following entry for each failed chkpnt:
> Mar  4 17:44:57 cluster-01 kernel: [174834.884946] CT: checkpointing
> not supported yet for hidden pid namespaces.
>
> I found that the following kernel parameter caused the issue:
> kernel.pid_ns_hide_child=1

That's it. Feel free to file a bug report for this missing feature, or 
write a small wiki article
to tell this to others.

In fact I just did that at 
https://openvz.org/Processes_scope_and_visibility#Hide_container_processes_from_host_completely
If there are any other places like this -- feel free to edit those.

>
> Too bad they don't work together.  It is probably worth noting on the
> wiki pages or in the man pages.
>
> Here is the output from a successful live migration:
>
> root at cluster-01:~# vzmigrate -s -v -t --live cluster-02 5000
> Starting live migration of CT 5000 to cluster-02
> OpenVZ is running...
>     Loading /etc/vz/vz.conf and /etc/vz/conf/5000.conf files
>     Check if ploop is supported on destination node
> Next unused minor: 1017904
>     Check IPs on destination node:
> Preparing remote node
>     Copying config file
> 5000.conf
>                                100% 1570     1.5KB/s   00:00
> Name fs01 assigned
> No changes in CT configuration, not saving
>     Creating remote container root dir
>     Creating remote container private dir
> Creating a container snapshot
> Creating snapshot {33d65014-8946-487c-93cc-4f17f4bd2527}
> Storing /vz/private/5000/Snapshots.xml.tmp
> Setting up checkpoint...
>          suspend...
>          get context...
> Checkpointing completed successfully
> Storing /vz/private/5000/root.hdd/DiskDescriptor.xml.tmp
> Creating delta /vz/private/5000/root.hdd/root.hdd.{f5e57811-b586-4e06-8957-3d63ca6d7c47}
> bs=2048 size=4614144 sectors
> Creating snapshot dev=/dev/ploop47831
> img=/vz/private/5000/root.hdd/root.hdd.{f5e57811-b586-4e06-8957-3d63ca6d7c47}
> ploop snapshot {33d65014-8946-487c-93cc-4f17f4bd2527} has been
> successfully created
> Setting up checkpoint...
>          join context..
>          dump...
> Checkpointing completed successfully
> Resuming...
> Snapshot {33d65014-8946-487c-93cc-4f17f4bd2527} has been successfully created
> Syncing private
> Live migrating container...
>     Copying top ploop delta with CT suspend
> Sending /vz/private/5000/root.hdd/root.hdd.{f5e57811-b586-4e06-8957-3d63ca6d7c47}
> Setting up checkpoint...
>          suspend...
>          get context...
> Checkpointing completed successfully
>     Dumping container
> Setting up checkpoint...
>          join context..
>          dump...
> Checkpointing completed successfully
>     Copying dumpfile
> dump.5000
>                                100% 8734KB   8.5MB/s   00:00
>     Undumping container
> Restoring container ...
> Error in ploop_fsck (fsck_util.c:376): Dirty flag is set
> Adding delta dev=/dev/ploop63619 img=/vz/private/5000/root.hdd/root.hdd (ro)
> Adding delta dev=/dev/ploop63619
> img=/vz/private/5000/root.hdd/root.hdd.{1cfb54ce-bade-4ca4-8882-ad322718974e}
> (ro)
> Adding delta dev=/dev/ploop63619
> img=/vz/private/5000/root.hdd/root.hdd.{f5e57811-b586-4e06-8957-3d63ca6d7c47}
> (rw)
> Mounting /dev/ploop63619p1 at /vz/root/5000 fstype=ext4 data='balloon_ino=12,'
> Container is mounted
>          undump...
> Setting CPU units: 1000
> Configure veth devices: veth5000.20 veth5000.40
> Adding interface veth5000.20 to bridge vmbr20 on CT0 for CT5000
> Adding interface veth5000.40 to bridge vmbr40 on CT0 for CT5000
>          get context...
> Container start in progress...
> Restoring completed successfully
>     Resuming container
> Resuming...
>
>          Suspend + Dump:   0.42
>     Pcopy after suspend:   0.50
>          Copy dump file:   0.84
>         Undump + Resume:   5.98
>                          ------
>    Total suspended time:   7.75
>
> Cleanup
>     Killing container
> Killing...
> Unmounting file system at /vz/root/5000
> Unmounting device /dev/ploop47831
> Container is unmounted
>     Removing dumpfiles
>     Destroying container
> Destroying container private area: /vz/private/5000
> Container private area was destroyed
>
> Axton Grams
> _______________________________________________
> Users mailing list
> Users at openvz.org
> https://lists.openvz.org/mailman/listinfo/users



More information about the Users mailing list