VMware plugin backups hang when using NBD mode and snapshots cannot be removed. (4040256)

VMware plugin backups hang when using NBD mode and snapshots cannot be removed.

When using the plugin for VMware, the backups may hang when NBD mode is used. 

The default read block size is set to 65536 in Plugin for VMware version 10.0.5. When multiple backups or streams are used and this large block size is used, the ESXi host can run out of NFC memory. When this happens the backup hangs and the disk handle is left in a valid state that can prevent snapshots from being consolidated or removed until the VM is rebooted. 

WORKAROUND

In the NetVault Backup web interface, go to Change Settings for the computer with the VMware Plugin installed. Under Plugin Options in the Plugin for VMware section adjust the "Read Block Size" to 2048. 

Plugin for VMware 10.5.0 will include VDDK 6.0 and will automatically retry with the smaller block size, if the ESXi host is also version 6.0+. 

Also see "Error returned when ESXi host reaches NFC memory limit" in the VMware VDDK 6.0 release notes.

https://www.vmware.com/support/developer/vddk/vddk-600-releasenotes.html

STATUS

Issue fixed in NetVault Backup plugin for VMware 10.5, but also requires that the ESXi host be updated with the fix for "Error returned when ESXi host reaches NFC memory limit." shown in the VDDK 6.0 release notes above. 

28780

Job may also fail with: Failed to read data from virtual disk with subtext: VDDK :01432 100 0 94041 Disk '[datastore] xxxxx_.vmdk' opened using mode 'nbd' VDDK :01432 19 0 94042 VDDK: [NFC ERROR] NfcFssrvrProcessErrorMsg: received diskLib error 2490377 from server: NFC_DISKLIB_ERROR VDDK :01432 45 0 94042 Failed to read disk data VDDK :01432 138 0 94042 VDDK error encountered: code 1 VDDK :01432 139 0 94042 Retrieving error message text VDDK :01432 52 0 94042 VDDK error message is 'NBD_ERR_DISKLIB' VMWSLAVE:01432 133 0 94042 Failed to read virtual disk at sector 0