[one-users] Problem (live)migrating VMs...

[Gonçalo Borges]

Hi Ruben et al.

I'm getting back to this issue, and as I've told previously in this 
thread, I'm able to perform live migrations. Nevertheless, the problems 
with a simple save/restore persist. I've started a VM using using 
OpenNebula, and in the cluster node where the machine is running, and 
using the oneadmin account, I can  save and restart the machine, if I 
save the checkpoint file in /tmp. Check following outputs:

---*---
-bash-3.2$ sudo /usr/sbin/xm list
Name                                      ID Mem(MiB) VCPUs State   Time(s)
Domain-0                                   0     4096     8 r-----   1665.1
one-4                                      6     1023     1 -b----     10.1

-bash-3.2$ sudo /usr/sbin/xm save 6 /tmp/6

-bash-3.2$ sudo /usr/sbin/xm list
Name                                      ID Mem(MiB) VCPUs State   Time(s)
Domain-0                                   0     4096     8 r-----   1674.5
-bash-3.2$ ll /tmp/6
-rwxr-xr-x 1 root root 1075387396 Aug 21 11:29 /tmp/6

-bash-3.2$ sudo /usr/sbin/xm restore /tmp/6

-bash-3.2$ sudo /usr/sbin/xm list
Name                                      ID Mem(MiB) VCPUs State   Time(s)
Domain-0                                   0     4096     8 r-----   1680.6
one-4                                      7     1023     1 -b----      0.0
---*---

Nevertheless, when I try to perform the same task in my predefined 
"VM_DIR=/srv01/cloud/images", it fails, even after setting the same 
permissions to this dir has /tmp. The problem here is that the 
checkpoint file is created but the machine continues to be listed after 
a xm list:


---*---
-bash-3.2$ ll /srv01/cloud/
total 0
drwxr-xr-x  3 oneadmin cloud 3896 Aug 19 12:44 conf
drwxrwxrwt  5 root     root  3896 Aug 21 11:34 images
drwxr-xr-x 10 oneadmin cloud 3896 Aug 19 12:32 one
drwxr-xr-x  6 oneadmin cloud 3896 Aug 11 18:31 xen

-bash-3.2$ sudo /usr/sbin/xm save 7 /srv01/cloud/images/7.chkpt

-bash-3.2$ ll /srv01/cloud/images/7.chkpt
-rwxr-xr-x 1 root root 13247 Aug 21 11:34 /srv01/cloud/images/7.chkpt

-bash-3.2$ sudo /usr/sbin/xm list
Name                                      ID Mem(MiB) VCPUs State   Time(s)
Domain-0                                   0     4096     8 r-----   1683.8
one-4                                      7     1023     1 ---s--      0.1
---*---

I do not understand what is the real difference between the two cases...

Cheers
Goncalo

>> * Migration / Save - Restore.
>>
>> It seems there is a problem with Xen restoring the images. Could you
>> check the following:
>>
>> 1.- Start a VM through OpenNebula
>> 2.- Go to the node where the VM is running and execute:
>> sudo /usr/sbin/xm save one-<VM_ID>  
>> /srv01/cloud/images/<VM_ID>/images/checkpoint
>> 3.- Check that the checkpoint file is created and its ownership
>> 4.- Restore the VM
>> sudo /usr/sbin/xm restore /srv01/cloud/images/<VM_ID>/images/checkpoint
>>
>> Check that you can do the save/restore with the oneadmin account, if
>> you have problems sometimes the xen log files can be useful...