...
Document Version Release Date Details 2 14 September 2017 Updated the Scope section and clarified in which version the fix can be found in the Resolution section 1 29 August 2017 Original Document Release HPE 3PAR Storage Replication Adapter (SRA) Software for VMware Site Recovery Manager (SRM) integrates HPE 3PAR StoreServ Storage with VMware SRM to provide a foundation for disaster recovery and business continuity solutions. When VMware Site Recovery Manager is configured with a stretched storage configuration, the primary and secondary volumes of a remote copy group may become writeable on both the protected and recovery site ESXi hosts when the following sequence of events occur: A Disaster Recovery is performed in a protected site array down or remote copy link down scenario, and the Virtual Machines (VM) are migrated to the recovery site ESXi host. When the protected site array comes up or remote copy link is UP, the replication is initiated in reverse direction using the Reprotect operation. At this stage, the new protected site array goes down. In order to migrate the VMs again, Disaster Recovery is performed. Later, when the array comes back due to an underlying SRA issue, the volumes will remain in read-write mode instead of read-only mode in the recovery site hosts. In this scenario, if a host application writes to the volumes on the recovery site, it might lead to data inconsistency as the volumes on both the protected and recovery site hosts are in read-write mode. As of the release date of this advisory, this issue has not been reported by any customer and has only been experienced in a lab environment.
This document applies to HPE 3PAR StoreServ arrays running HPE 3PAR OS version 3.2.2 MU4 or 3.3.1 with HPE 3PAR SRA version 6.1, 6.5, or 6.5.1 in VMware SRM stretched storage configuration.
This issue is addressed in HPE 3PAR SRA version 6.5.2. HPE 3PAR StoreServ arrays running HPE 3PAR OS version 3.2.2 MU4 should be upgraded to version 3.3.1 before installing HPE 3PAR SRA 6.5.2. Workaround When the customer ends up in a scenario where the new protected site array goes down (step 3 in the scenario above), perform the following steps before executing Disaster Recovery: Before performing the disaster recovery operation, remove the I/O paths between recovery site ESXi hosts and to the array which is currently down so that no writes are propagated to the devices once the array comes back. Perform the VMware SRM disaster recovery operation. Once the disaster recovery operation is complete, the VMware SRM displays the Recovery Required prompt. This operation implies that after bringing up the storage array or replication link, the recovery operation must be executed. Once the array is up, execute the following HPE 3PAR CLI commands: Execute showrcopy on both the arrays and verify that that the remote copy group role is Primary on the failed array that has come back and the role is Primary-Rev in the array where the VMs are currently online. Execute the following command in the array where VMs are online and the remote copy group role is Primary-Rev : setrcopygroup recover -nostart <remotecopygroupname> Execute the showrcopy command on the array where the group role was Primary. Ensure that the remote copy group rols is now changed from from Primary to Secondary-Rev. Restore the I/O paths between the ESXi hosts and the array. Reboot the ESXi hosts where the VMs were online earlier. Navigate to the Devices tab in the VMware SRM and click Refresh to rediscover the devices. Make sure that the devices are discovered again in the VMware SRM. Execute the Recovery required operation followed by the Reprotect operation. Contact HPE Support if you face any issues while executing these steps. RECEIVE PROACTIVE UPDATES : Receive support alerts (such as Customer Advisories), as well as updates on drivers, software, firmware, and customer replaceable components, proactively in your e-mail through HPE Subscriber's Choice. Sign up for Subscriber's Choice at the following URL: Proactive Updates Subscription Form