Skip to main content

Hi Vaulters,

 

Our customer had some concerns regarding two DR scenarios, in which your guidance would be much appreciated.

 

Below both scenarios :

  1. In case of loss of a MediaAgent, how can the data be restored from the storage bay ? Do we have to install a MediaAgent from Scratch ? Knowing that we have a second MediaAgent. In the same case, how about DDB, can it be rebuilt from the Storage Bay backup images ?
  2. Our customer also has secondary copies for each storage policy on Tape, how can they restore the data in case of a complete loss of the Infra (Only Tape library remaining) ? Is there any way to rebuild the CS from the tape ? Same for the MediaAgents ?

We heard about Commvault’s Media Explorer, but didn’t get what it does exactly, and if it is the main solution for rebuilding the CS et MAs metadata, is it compatible for Tape also ?

 

Kind regards


To perform restores in the event of DR, you need the CS, a MA and access to the storage.  It’s best to make sure all of those components are going to be available in case of a DR, instead of waiting for an actual event and building them from scratch.

Thanks,
Scott
 



To perform restores in the event of DR, you need the CS, a MA and access to the storage.  It’s best to make sure all of those components are going to be available in case of a DR, instead of waiting for an actual event and building them from scratch.

Thanks,
Scott
 

Hi Scott,

 

Thanks for your valuable feedback.

 

As per our understanding, it is possible to rebuild the CS DB from tape using the MediaExplorer tool, is that true ?

After rebuilding the CS using the restored DB from Tape, how can we proceed with MA restore ? Prepare Hardware, install OS and CV binaries ? And how about Index Cache and DDBs ?

 

Kind regards.


Do NOT rely on Media Explorer to save your CSDB.  There are SEVERAL ways to protect your CSDB; and you should be using 1 or more of them.

Configuration of Disaster Recovery (DR) Backups
https://documentation.commvault.com/2023e/essential/configuration_of_disaster_recovery_dr_backups.html

If you don’t have DR MA on standby, yes, you will need to deploy HW, the OS and install CV.  You don’t need to worry about the Index Cache and DDB, they will get automatically recovered from the backup media.

However, as I mentioned before, the BEST approach here is to have the DR CS and a DR MA already deployed and ready.  Make your DR situation easier by being ready to go instead of adding more stress by needing to rebuild infrastructure.

Thanks,
Scott
 


Thank you, @Scott Moseman, for your prompt feedback.

 

Yes, having a standby MA is the best approach to go with. 

 

Regarding the fact that in case of we have to fully restore the MA, how does the process of recovering the DDB and Index Cache works ? I’d love if you can enlighten me.

 

Thanks.


That’s why we have DDB and Index Cache backups.

Recovering Permanently Offline DDB Partitions
https://documentation.commvault.com/2023e/expert/recovering_permanently_offline_ddb_partitions.html

I believe the Index Cache automatically rebuilds itself when necessary from the Index backups.
Often times sealing the DDB and starting over is quicker & easier than trying to rebuild the DDB.

Thanks,
Scott


Hi Scott,

 

We had one more concern, as we mentioned it before, in our main DC, we have two MAs, having both access to the same SAN Disk Library.

Each MA has its proper LUNs that are mapped to it (zonning) with each LUN being configured as a mount path on CV.

 

In case we lose one of the MAs, is there any way to map the LUNs of the lost MA directly to the second one ? And somehow being able to restore the data backed up by the lost MA using the remaining MA ?

 

Kind regards.


I will let anyone else chime in, but I don’t think you can re-map SAN LUNs and still use the data.  This is why I suggest people to use NAS because it’s super easy to load balance, share, etc.

Thanks,
Scott
 


I will let anyone else chime in, but I don’t think you can re-map SAN LUNs and still use the data.  This is why I suggest people to use NAS because it’s super easy to load balance, share, etc.

Thanks,
Scott
 

Hi Scott, 

 

Thanks for your prompt feedback. We would have loved to be able to implement everything through NAS, as this would simplify a lot of things regarding MA HA, but unfortunately, our Infra is mainly SAN based.

 

We came across this link : https://documentation.commvault.com/11.24/expert/10941_change_data_path_overview.html

 

Is this any helpful in our case ? Thanks in advance.

The main goal to achieve, is to be able to tell Commvault, that the mount path still exists (even if the original MA is not available), but, it is now under another path (Another MA, another partition …etc.), and the backed up data is the same and available, just being accessible by another MA. Which will give us the possibility to restore that data using the other MA.

 

Regards.


I’m not sure if that process is doing to do what you want or not.  I would suggest setting up a test library and storage policy and giving it a try.  I see it mentions using Move Mount Path for disk libraries, which I believe requires the source to be online to copy over the data.  This definitely should be tested to confirm it works as you expect before assuming it’ll work in a DR situation.

Thanks,
Scott
 


I’m not sure if that process is doing to do what you want or not.  I would suggest setting up a test library and storage policy and giving it a try.  I see it mentions using Move Mount Path for disk libraries, which I believe requires the source to be online to copy over the data.  This definitely should be tested to confirm it works as you expect before assuming it’ll work in a DR situation.

Thanks,
Scott
 

Hi Scott,

 

Just to let you know, and other Vaulters, that it is feasible through SAN as we tested it.

 

  1. Map Mount Path LUNs to other second MA (Storage Bay mapping + Zoning).
  2. Scan LUNs on second MA.
  3. Share Mount Path from MA1 to MA2 through Mount Path sharing on Commvault while specifying the local partition of the LUNs on the OS of the second MA.
  4. Test restore → working.

Thanks again for your support.

 

Regards.

 


Reply