If you use Site Recovery Manager to test or recover a large number of virtual machines, you might need to modify the default Site Recovery Manager settings to achieve the best possible recovery times in your environment or to avoid timeouts.

In large environments, Site Recovery Manager might simultaneously power on or power off large numbers of virtual machines. Simultaneously powering on or powering off large numbers of virtual machines can create a heavy load on the virtual infrastructure, which might lead to timeouts. You can modify certain Site Recovery Manager settings to avoid timeouts, either by limiting the number of power on or power off operations that Site Recovery Manager performs concurrently, or by increasing the timeout periods.

The limits that you set on power on or power off operations depend on how many concurrent power on or power off operations your infrastructure can handle.

You modify certain options in the Advanced Settings menus in the vSphere Web Client or in the Site Recovery Manager client plug-in. To modify other settings, you edit the vmware-dr.xml configuration file on the Site Recovery Manager Server. Always modify settings by using the client menus when an option exists. If you modify settings, you must make the same modifications on the Site Recovery Manager Server and vCenter Server instances on both the protected and recovery sites.

For descriptions of the settings that you can change, see Settings for Large Site Recovery Manager Environments.

Procedure

  1. In the vSphere Web Client, select a cluster.
  2. On the Manage tab, select Settings > vSphere DRS.
  3. Click Edit.
  4. In Advanced Options, set the srmMaxBootShutdownOps setting.
    Option Description
    Option text box Enter srmMaxBootShutdownOps.
    Value text box Enter the maximum number of boot and shutdown operations, for example 32. If you set the value to 32, the next guest starts booting or shutting down as soon as one of the first batch of 32 has finished, namely VMs 1 to 32 all start together, then VM 33 starts once one of the first batch has finished, VM 34 starts when the second one of the first batch has finished, and so on.
  5. Click OK to save your changes.
  6. Log in to the Site Recovery Manager Server host.
  7. Open the vmware-dr.xml file in a text editor.
    You find the vmware-dr.xml file in the C:\Program Files\VMware\VMware vCenter Site Recovery Manager\config folder.
  8. Change the defaultMaxBootAndShutdownOpsPerCluster and defaultMaxBootAndShutdownOpsPerHost settings in the vmware-dr.xml file: 
    <config>
...
   <defaultMaxBootAndShutdownOpsPerCluster>24</defaultMaxBootAndShutdownOpsPerCluster>
   <defaultMaxBootAndShutdownOpsPerHost>4</defaultMaxBootAndShutdownOpsPerHost>
...
</config>
    If these elements do not already exist in the vmware-dr.xml file, you can add them anywhere in the <config> section. If you set the <defaultMaxBootAndShutdownOpsPerCluster> value to 24, the next guest starts booting or shutting down as soon as one of the first batch of 24 has finished, namely VMs 1 to 24 all start together, then VM 25 starts once one of the first batch has finished, VM 26 starts when the second one of the first batch has finished, and so on.
  9. Restart the Site Recovery Manager Server to apply the new settings.
  10. In the vSphere Web Client, click Site Recovery > Sites, and select a site.
  11. Select Advanced Settings > vSphere Replication and increase the vrReplication.synchronizationTimeout setting.
    The default value is 7200 and corresponds to a working synchronization timeout period of 14400 seconds.
  12. Select Advanced Settings > Storage and increase the storage.commandTimeout setting.
    The default value is 300 seconds.
  13. Click OK to save your changes.