Symptoms
The issue stems from a Zerto bug when the Nutanix IO Control feature is exercised. Nutanix IO Control can be triggered when IO control is transferred to another node for high availability, including during routine Nutanix upgrades. When this transfer occurs, there can be a brief period where Zerto incurs IO errors. These IO errors are different from regular IO errors, as Nutanix is using a unique error code which may be handled incorrectly by Zerto. The Nutanix environment needs to be in a certain state for the Zerto zdriver to misinterpret these IO status messages. If this occurs, data corruption could occur in protected virtual machines on the original node. It is important to note that ZVR is not corrupting the data, rather the ZVR was incorrectly handling the failure code.
Solution
Zerto released ZVR version 5.0u2, containing host components that fix the handling of the error conditions. In the case where customers can not move from 4.5, Zerto will release 4.5u5 (patch 1) containing the host component VRA for vmware, tentatively on March 2, 2017.
- NOTE: Upgrading to 5.0u2 from any 4.5 version will incur a delta sync
- NOTE: Upgrading to 5.0u2 from any 5.0 version will NOT incur a delta sync
- NOTE: Upgrading to 4.5u5 patch 1 from any previous version of ZVR will incur a delta sync