We run a vRanger 6.0.1 nightly job to back up approximately 180 VMs residing across 4 vSphere 5.1 hosts. Every morning there are several VMs in vCenter that show up with the "virtual machine disks consolidation is needed" yellow exclamation warning on them. It is not always the same VMs, just a random handful every night. Manual consolidation is needed on them as the delta files are left on the VMs. The vRanger task log for a problem VM indicates no problems, "Removing snapshot for vRanger completed." There isn't a single error throughout the task log. So from the vRanger standpoint, it thinks it completed correctly. I assume this is because of poor error handling however as if you look at the events for the VM in question in vCenter, coinciding at the same time as vRanger removing the snapshot line in its log, you see events for "Task: Remove snapshot" followed within seconds by "Virtual machine disks consolidation is needed" followed immediately by "Virtual machine disks consolidation failed." So the reference to the snapshot is being removed but consolidation fails leaving the snapshot in an orphanned ghosted state and growing until manual consolidation is performed.
All of these VMs that experience this problem are using Machine-based VDDK SAN "HotAdd" so I am assuming that the physical server that runs vRanger is not letting go of the VMDK before consolidation is attempted so the VMDK lock is preventing the consolidation from occurring. The SANs in question are a pair of EqualLogics connected via iSCSI to both the vSphere hosts and the physical server that runs vRanger.
Any ideas how to prevent this from happening?