Hi @FeRox,

There could be a number of things causing this to occur, so we’ll need more details here.

-Are both Hyper-V nodes added as “Nodes” under the VirtualServer Instance Properties?
-How many VM’s are configured in the subclient’s?
-How many “Data Readers” (Streams) are configured in the subclient(s) Properties?
-Are the VM’s distributed between Nodes and CSV’s?

In the vsbkp.log on the node (being used for backups), you will see the lines stating “Adding VM VMNAME] on AHOST] to Agent ENODENAME] ]REASON]

-This will give an indication of what rule was used to dispatch the VM to that specific VirtualServer Agent.

Best Regards,

Michael

Hi I hope you are well, I answer the questions.

-Are both Hyper-V nodes added as "Nodes" in the Virtual Server Instance Properties?
Yes.

-How many virtual machines are configured on the subclient?
60 virtual machines, 32 on one and 28 on the other node.

-How many "Data Readers" (Streams) are configured in the subclient properties?
it has 5 data readers

-Are the VM's distributed between Nodes and CSV's?
yes, 32 in one node and 28 in the other node
 
In the vsbkp.log of the node (which is used for backups), you will see the lines that say "Add VM VMNAME] on HOST] to Agent NODENAME] REASON]".

I attach the log, but I don't see the node being used, however on the process tab you only see the use of one mediaagent.

-This will give an indication of what rule was used to send the VM to that specific VirtualServer Agent.

Thanks @FeRox,

In theory if both Nodes are added and the Subclient has 5 streams set, we should be able to leverage both nodes for the backups and split the streams between those. (Depending on available resources and stream configurations).

The log provided is only partial here, We may need the entire log file here to check further. 

Best Regards,
Michael

please could you tell me how can I get the whole log file to attach it?

@FeRox , the easiest way is to right click the job or client and select Send Logs.  Save them locally and attach the vsbkp.log file from the package here.

Thank you very much mike, here is the link to download the logs, thank you very much for your help, they are very heavy and I can not attach them.

Donwload_Logs_Clic_Here.

Hi @FeRox,

I had a quick look over the log attached. - For the Majority of the VM’s we’re using “CSV Owner” dispatch meaning that the Node owning the CVS is backing up the VM’s on the CSV.

Since one node is hosting more of the CSV’s, this is why you see an imbalance:

20980 f64   10/07 07:12:55 24545 VSBkpCoordinator::DistributeVMsToAgents() - Agent nBOA02SRVSH121] has 33 VMs (33 static)  57 max streams
20980 f64   10/07 07:12:55 24545 VSBkpCoordinator::DistributeVMsToAgents() - Agent gBOA02SRVSH122] has 4 VMs (4 static)  8 max streams
 

If you wish to split the load more event between hosts, I would suggest Create the bEnableCSVOwnerDispatch additional setting on the VSA proxy and set the value to false. - This should then use the Host Dispatch method.

Here’s the documentation for applying Additional Settings: https://documentation.commvault.com/2022e/expert/8681_adding_additional_setting_from_commcell_console.html

More details on the performance tuning settings for VM backups can be found here: https://documentation.commvault.com/2022e/expert/31115_performance_tuning_for_backups.html

Best Regards,

Michael

Good afternoon Mike, I have made the configurations recommended by you in another Backup with the same characteristics, but it still takes the same MediaAgent (cca00srvssy02), please note that all HOSTS have the MediaAgent role, and what I want is that in the MediaAgents column it takes the 4 available servers and not just one.