How to re-attach FC storage after reinstalling rhev manager ?

Latest response

Hi,

After successfull setup of a rhevm with 2 rhevh both connected to an NFS ISO and a FC domain, I decided to reinstall the manager from scratch (with the last version as the previous was beta one).

 

I managed reattaching the NFS ISO but I'm unable to reattach FC storage.

 

This is the Multipath -ll output showing that both LUNs are available:

3600508b40006bd730005300000090000 dm-1 HP,HSV200
size=150G features='1 queue_if_no_path' hwhandler='0' wp=rw
|-+- policy='round-robin 0' prio=50 status=active
| |- 0:0:1:1 sdc        8:32  active ready running
| `- 1:0:0:1 sde        8:64  active ready running
`-+- policy='round-robin 0' prio=10 status=enabled
  |- 0:0:0:1 sda        8:0   active ready running
  `- 1:0:1:1 sdg        8:96  active ready running

3600508b40006bd730005300000100000 dm-2 HP,HSV200
size=151G features='1 queue_if_no_path' hwhandler='0' wp=rw
|-+- policy='round-robin 0' prio=50 status=active
| |- 0:0:1:2 sdd        8:48  active ready running
| `- 1:0:0:2 sdf        8:80  active ready running
`-+- policy='round-robin 0' prio=10 status=enabled
  |- 0:0:0:2 sdb        8:16  active ready running
  `- 1:0:1:2 sdh        8:112 active ready running
[root@rhevh2 ~]#

I successfully managed creating a new FC domain using my second LUN (151GB one).

 

Is there any thing to do to get this storage back up?

 

Regards

Bernard
 

BM
Log in to join the conversation

Responses

Sadique Puthen's picture Red Hat Guru 4679 points
6 March 2012 9:31 AM

Hi,

 

If you are looking to create a new Storage domain using the lun and don't want to preserve guest images from old set up or other data that may be in the lun, then you need to wipe out the data from the lun using dd.

 

dd if=/dev/zero of=/dev/lun/lun

 

Then press ctr + c after 5 seconds and try to re-add the storage domain.

 

If there are guest images in the lun that need to be retrieved, there is no supported way to do that if you have not exported them to an export domain.

BM Active Contributor 378 points
6 March 2012 2:14 PM

Thank you. I expected being able to recover some vitual machines!

I don't loose any critucal data (just precious time!) and will reinstall those VMs.

So a question would be: what if a production RHEVM manager fail?

Is there any conservative things to do (exports?) so that to be able to recover

VMs in case of?

DY Guru 5340 points
6 March 2012 2:27 PM

To protect against RHEV-M failure, you need to have a reecent backup of the RHEV-M database. With the database and the LUNs intact, you can recover from any failure.

 

In production I normally dump my database using a simple script once every hour, and rsync those dump files to a remote host. This is something VERY easily done, through just a couple of lines in bash and cron.

 

You need to understand that the storage domain only contains VM images, with no indication as to which image belongs to which VM, and which belongs to a template and not a VM. This is why there are export domains by the way - during the export, the VM data is dumped into an OVF file that is stored on the export domain along with the images

 

 

The database contains the data that maps VM disk images to VMs, snpashots and templates. These two need to work together, and without the database, the storage domain is just a bunch of images, where it's not easy to determine which used to belong to which VM.

 

Sadique Puthen's picture Red Hat Guru 4679 points
6 March 2012 2:28 PM

> what if a production RHEVM manager fail?

 

If you have database backup from the crashed rhevm, it's possible to restore the database to the newly installed RHEVM so that you can use everything from the old storage domain. Make it a practice to backup database and other configuration files regularly.

 

> Is there any conservative things to do (exports?) so that to be able to recover

VMs in case of?

 

Ofcourse, you need to backup data in your vms. You can regularly backup your vms to export domain or you can run a backup agent inside you vm to take backup of important files and application to a remote server. So if the vm crashes beyond recovery, you can import from the export domain or provision a fresh vm and restore application data from backup.

ka Active Contributor 150 points
7 June 2012 4:57 PM

it is  posible  if   you   have  a export  domain  and  luns are  intact :) ,  the procedure is odd but  it could be  possible ,  

 

the  requeriments  are

 

1)  Reinstall  the  RHEV-M,  configure  a  Datacenter , Cluster, Hosts,  Storage Data Domain(FC/iSCSI)  active

2)  Configure  a  export  domain with suficient space to  exports  vms.

3) Connect  the  luns to RHEV-M host

4) scan   your  disk  to search vgs :

   4.1) lvmdiskscan

   4.2) pvscan

   4.3) vgscan

   4.4) vgchange  -ay  vg_name

 

5)  look  to lvs  master  ,   mount  /dev/vgname/master  /mnt/rhevmaster

6)  cd  vms,  you will see  a  vm ids   dirs

7)  cd  eb8aab1a-e94b-4d89-9ca4-de6db1610d74

8)  view file.ovf 

9) look  for ids of  disks  virtual  machine   grep --color -i  disk  file.ovf

10) create a  virtual machine  importing  disks using virt-manager

11) Test the  machine  run

12) Shutdown  the  vm

13) Active  and  Attach  the  Export Storage  Domain

14) Use  virt-v2v   to export the  vm again  to RHEV in  new Data  Storage Domain

 

kind regards,