Non-disruptive storage device migration in failover cluster environment

• US 8,775,861 B1
• Filed: 06/28/2012
• Issued: 07/08/2014
• Est. Priority Date: 06/28/2012
• Status: Active Grant

• Current Assignee: Emc IP Holding Company LLC (Dell Technologies Inc.)
• Original Assignee: EMC Corporation (Dell Technologies Inc.)
• Inventors: Raizen, Helen S., Contreras, Cesareo, Bappe, Michael E., Thigpen, Edward L.
• Primary Examiner(s): DUNCAN, MARC M
• Application Number:
  US13/535,740
• Time in Patent Office: 740 Days
• Field of Search: 714/4.11, 714/4.12, 714/6.3, 714/6.32, 714/16
• US Class Current: 714/4.12
• CPC Class Codes: G06F 11/1662 the resynchronized component or unit being a persistent storage device re-synchronization of failed mirror storage G06F11/2082; rebuild or reconstruction of parity RAID storage G06F11/1008
• CPC Class Codes: G06F 11/2028 eliminating a faulty processor or activating a spare
• CPC Class Codes: G06F 11/2094 Redundant storage or storage space G06F11/2056 takes precedence
• CPC Class Codes: G06F 3/0647 Migration mechanisms

Abstract

A method of performing data migration from a source storage device to a target storage device in a failover cluster includes use of a roll-forward flag to signal successful completion of a migration operation from a migration node to failover nodes of the cluster, reliably controlling host access to the target storage device to ensure that it is used only when it has been successfully synchronized to the source storage device and a commit operation has occurred that ensures that subsequent read and write operations are directed exclusively to the target storage device.

Claims

1. A method of non-disruptively migrating contents of a source storage device to a target storage device in a data processing system having a set of host computers organized into a failover cluster, the host computers including a migration node and one or more failover nodes, the migration node executing an application program accessing the source storage device via a device name, each failover node being operative to begin execution of the application program upon failure of the migration node, comprising:

creating metadata and a roll-forward flag, the metadata indicating that a migration operation is in progress, the roll-forward flag initially being reset, the metadata also including access control data initially set to allow access to the source storage device and disallow access to the target storage device;

subsequently synchronizing the target storage device to the source storage device and configuring the target storage device to each of the failover nodes, the synchronizing leading to a source-selected operating state in which read and write operations continue to be directed to the source storage device and write operations are also duplicated to the target storage device; and

subsequently performing a commit operation causing subsequent read and write operations to be directed exclusively to the target storage device, the commit operation including (i) setting the roll-forward flag, (ii) setting the access control data to disallow access to the source storage device and allow access to the target storage device, and (iii) changing or remapping the device name to cause the application program to access the target storage device instead of the source storage device;

wherein the migration node is operative when the source storage device is coming online (i) if the roll-forward flag is not set, to abort the migration operation and maintain the initial setting of the access control data allowing access to the source storage device and disallowing access to the target storage device, and (ii) otherwise if the roll-forward flag is set, to ensure completion of the commit operation;

and wherein each failover node is operative when the source storage device is coming online and the roll-forward flag is set, to (i) set the access control data to disallow access to the source storage device and allow access to the target storage device, and (ii) change or remap the device name to cause an application program executing on the failover node to access the target storage device instead of the source storage device.

2. A method according to claim 1, wherein each host includes a filter driver and a set of clustering components, the filter driver including a kernel-level part performing a privileged part of a migration tool used to control the migrating operation, the cluster component including a migration-aware cluster component executed prior to application use of the source storage device, and wherein the metadata includes kernel metadata accessed by the kernel-level part of the filter driver and the migration component to set and test the value of the access control data.

3. A method according to claim 2, wherein the migration node further includes a user-level part of the migration tool having access to user metadata used to persistently track progress of the migration operation across restarts of the migration host.

4. A method according to claim 1, further including configuring the target storage device to the failover nodes at a time prior to the commit operation when the access control data disallows host access to the target storage device.

5. A method according to claim 4, further including explicitly initiating a reading of the access control data by each failover node prior to configuring the target storage device to the failover nodes to ensure each of the failover nodes is prevented from accessing the target storage device.

6. A method according to claim 1, wherein the commit operation is commenced from the source selected operating state.

7. A method according to claim 1, further including a cleanup operation subsequent to the commit operation, the cleanup operation including unconfiguring the source storage device from each host computer and removing the roll-forward flag.

8. A computerized device operable as a migration node in a set of host computers organized into a failover cluster, the host computers also including one or more failover nodes, the migration node executing an application program accessing a source storage device via a device name, each failover node being operative to begin execution of the application program upon failure of the migration node, comprising:

memory storing computer program instructions;

one or more processors;

input/output interface circuitry for connecting the computerized device to the other host computers and to the source storage device and to a target storage device; and

interconnect circuitry coupling the memory, processors and input/output interface circuitry together,

wherein the processors are operative to execute the computer program instructions from the memory to cause the computerized device to non-disruptively migrate contents of the source storage device to the target storage device by;

creating metadata and a roll-forward flag, the metadata indicating that a migration operation is in progress, the roll-forward flag initially being reset, the metadata also including access control data initially set to allow access to the source storage device and disallow access to the target storage device;

subsequently synchronizing the target storage device to the source storage device, the synchronizing leading to a source-selected operating state in which read and write operations continue to be directed to the source storage device and write operations are also duplicated to the target storage device; and

subsequently performing a commit operation causing subsequent read and write operations to be directed exclusively to the target storage device, the commit operation including (i) setting the roll-forward flag, (ii) setting the access control data to disallow access to the source storage device and allow access to the target storage device, and (iii) changing or remapping the device name to cause the application program to access the target storage device instead of the source storage device;

wherein the migration node is operative when the source storage device is coming online (i) if the roll-forward flag is not set, to abort the migration operation and maintain the initial setting of the access control data allowing access to the source storage device and disallowing access to the target storage device, and (ii) otherwise if the roll-forward flag is set, to ensure completion of the commit operation;

and wherein each failover node is operative when the source storage device is coming online and the roll-forward flag is set, to (i) set the access control data to disallow access to the source storage device and allow access to the target storage device, and (ii) change or remap the device name to cause an application program executing on the failover node to access the target storage device instead of the source storage device.

9. A computerized device according to claim 8, wherein each host includes a filter driver and a set of clustering components, the filter driver including a kernel-level part performing a privileged part of a migration tool used to control the migrating operation, the cluster component including a migration-aware cluster component executed prior to application use of the source storage device, and wherein the metadata includes kernel metadata accessed by the kernel-level part of the filter driver and the migration component to set and test the value of the access control data.

10. A computerized device according to claim 9, wherein the migration node further includes a user-level part of the migration tool having access to user metadata used to persistently track progress of the migration operation across restarts of the migration host.

11. A computerized device according to claim 8, wherein the target storage device is configured to the failover nodes at a time prior to the commit operation when the access control data disallows host access to the target storage device.

12. A computerized device according to claim 11, wherein configuring the target device at each failover node is preceded by an explicit reading of the access control data at each failover node to ensure each of the failover nodes is prevented from accessing the target storage device.

13. A computerized device according to claim 8, wherein the commit operation is commenced from the source selected operating state.

14. A computerized device according to claim 8, wherein the commit operation is followed by a cleanup operation, the cleanup operation including unconfiguring the source storage device from each host computer and removing the roll-forward flag.