Difference between revisions of "Migration fails"

From OpenVZ Virtuozzo Containers Wiki
Jump to: navigation, search
m (spelling/wording)
(Solution)
 
(11 intermediate revisions by 5 users not shown)
Line 1: Line 1:
This article contents known reasons of migration failures.
+
This article describes known reasons of migration failures, and solutions for those.
  
 
== Problem #1 ==
 
== Problem #1 ==
=== Symptoms ===
+
=== Symptoms ===
 
After giving <tt>vzmigrate -v --online <ip address> <veid></tt>
 
After giving <tt>vzmigrate -v --online <ip address> <veid></tt>
 
the following message appears:
 
the following message appears:
 
<pre>
 
<pre>
Error: undump failed: Invalid argumentStopping VE ...
+
Error: undump failed: Invalid argument
 +
Stopping container ...
 
Restoring failed:
 
Restoring failed:
 
iptables-restore exited with 1
 
iptables-restore exited with 1
Line 13: Line 14:
  
 
=== Reason ===
 
=== Reason ===
The VE which you wish to migrate uses some iptables modules, that are not
+
The container which you wish to migrate uses some iptables modules, that are not
 
loaded on destination node. Consequently <tt>iptables-restore</tt> fails.
 
loaded on destination node. Consequently <tt>iptables-restore</tt> fails.
  
=== Workaround ===
+
=== Solution ===
Load all iptables kernel modules that are used by VE on destination node.
+
Load all iptables kernel modules that are used by container on destination node.
 +
Also, it seems like iptables needs to be installed inside the container.
  
 
=== Links ===
 
=== Links ===
  
* http://bugzilla.openvz.org/show_bug.cgi?id=631
+
* {{bug|631}}
 
* http://forum.openvz.org/index.php?t=tree&th=2756&mid=14198&&rev=&reveal=
 
* http://forum.openvz.org/index.php?t=tree&th=2756&mid=14198&&rev=&reveal=
  
 
== Problem #2 ==
 
== Problem #2 ==
=== Symptoms ===
+
=== Symptoms ===
 
The system runs <tt>028stab031</tt> or older kernels and  
 
The system runs <tt>028stab031</tt> or older kernels and  
 
after giving <tt>vzmigrate -v --online <ip address> <veid></tt>
 
after giving <tt>vzmigrate -v --online <ip address> <veid></tt>
 
the following message appears:
 
the following message appears:
 
<pre>
 
<pre>
Live migrating VE
+
Live migrating container
     Suspending VE
+
     Suspending container
 
Setting up checkpoint...
 
Setting up checkpoint...
 
         suspend...
 
         suspend...
Can not suspend VE: Device or resource busy
+
Can not suspend container: Device or resource busy
 
file on unsupported FS: magic 0bad1dea
 
file on unsupported FS: magic 0bad1dea
 
Checkpointing failed
 
Checkpointing failed
Error:  Failed to suspend VE
+
Error:  Failed to suspend container
 
</pre>
 
</pre>
  
 
=== Reason ===
 
=== Reason ===
The VE which you wish to migrate uses inotify mechanism (magic 0bad1dea), that is not
+
The container which you wish to migrate uses inotify mechanism (magic 0bad1dea), that is not
 
supported by <tt>028stab031</tt> and older kernels.
 
supported by <tt>028stab031</tt> and older kernels.
  
=== Workaround ===
+
=== Solution ===
Update till <tt>028stab035</tt> or newer kernel.
+
* for 2.6.18 rhel5 based kernels: Update till <tt>028stab035</tt> or newer kernel
 +
* for 2.6.18 mainstream based kernels: Update till <tt>028stab043</tt> or newer kernel
  
 
=== Links ===
 
=== Links ===
  
* http://bugzilla.openvz.org/show_bug.cgi?id=552
+
* {{bug|552}}
 
* http://forum.openvz.org/index.php?t=tree&goto=14780&#msg_14780
 
* http://forum.openvz.org/index.php?t=tree&goto=14780&#msg_14780
  
  
 
== Problem #3 ==
 
== Problem #3 ==
=== Symptoms ===
+
=== Symptoms ===
 
After giving <tt>vzmigrate -v --online <ip address> <veid></tt>
 
After giving <tt>vzmigrate -v --online <ip address> <veid></tt>
 
the following message appears:
 
the following message appears:
 
<pre>
 
<pre>
Dumping VE
+
Dumping container
 
Setting up checkpoint...
 
Setting up checkpoint...
 
         join context..
 
         join context..
 
         dump...
 
         dump...
Can not dump VE: Device or resource busy
+
Can not dump container: Device or resource busy
 
unsupported netdevice tun0
 
unsupported netdevice tun0
 
Checkpointing failed
 
Checkpointing failed
Error:  Failed to dump VE
+
Error:  Failed to dump container
 
</pre>
 
</pre>
  
 
=== Reason ===
 
=== Reason ===
Checkpointing doesn't support <tt>tun</tt> device inside VE, which is usually used  
+
Checkpointing doesn't support <tt>tun</tt> device inside container till <tt>2.6.18-028stab049</tt> and <tt>2.6.9-023stab044.14</tt> kernels. This device is usually used for [[VPN via the TUN/TAP device]] in container.
for [[VPN via the TUN/TAP device]] in VE.
 
  
=== Workaround ===
+
=== Solution ===
You should not use <tt>tun</tt> device in VE to migrate  it. If you're using VPN in VE, just stop VPN, then produce the online migration and after that start the VPN. Of course, all current connections to VPN will be lost :(.
+
* for 2.6.18 based kernels: update till <tt>028stab049</tt> or newer kernel
 +
* for 2.6.9 based kernels: update till <tt>023stab044.14</tt> or newer kernel
  
 
=== Links ===
 
=== Links ===
* http://bugzilla.openvz.org/show_bug.cgi?id=642
+
* {{Bug|642}}
 
* http://forum.openvz.org/index.php?t=tree&goto=14826&#msg_14826
 
* http://forum.openvz.org/index.php?t=tree&goto=14826&#msg_14826
  
 
[[Category:Troubleshooting]]
 
[[Category:Troubleshooting]]

Latest revision as of 11:39, 7 May 2010

This article describes known reasons of migration failures, and solutions for those.

Problem #1[edit]

Symptoms[edit]

After giving vzmigrate -v --online <ip address> <veid> the following message appears:

Error: undump failed: Invalid argument
Stopping container ...
Restoring failed:
iptables-restore exited with 1
rst_restore_net: -22

Reason[edit]

The container which you wish to migrate uses some iptables modules, that are not loaded on destination node. Consequently iptables-restore fails.

Solution[edit]

Load all iptables kernel modules that are used by container on destination node. Also, it seems like iptables needs to be installed inside the container.

Links[edit]

Problem #2[edit]

Symptoms[edit]

The system runs 028stab031 or older kernels and after giving vzmigrate -v --online <ip address> <veid> the following message appears:

Live migrating container
    Suspending container
Setting up checkpoint...
        suspend...
Can not suspend container: Device or resource busy
file on unsupported FS: magic 0bad1dea
Checkpointing failed
Error:  Failed to suspend container

Reason[edit]

The container which you wish to migrate uses inotify mechanism (magic 0bad1dea), that is not supported by 028stab031 and older kernels.

Solution[edit]

  • for 2.6.18 rhel5 based kernels: Update till 028stab035 or newer kernel
  • for 2.6.18 mainstream based kernels: Update till 028stab043 or newer kernel

Links[edit]


Problem #3[edit]

Symptoms[edit]

After giving vzmigrate -v --online <ip address> <veid> the following message appears:

Dumping container
Setting up checkpoint...
        join context..
        dump...
Can not dump container: Device or resource busy
unsupported netdevice tun0
Checkpointing failed
Error:  Failed to dump container

Reason[edit]

Checkpointing doesn't support tun device inside container till 2.6.18-028stab049 and 2.6.9-023stab044.14 kernels. This device is usually used for VPN via the TUN/TAP device in container.

Solution[edit]

  • for 2.6.18 based kernels: update till 028stab049 or newer kernel
  • for 2.6.9 based kernels: update till 023stab044.14 or newer kernel

Links[edit]