System and method for virtual interface failover within a cluster

• US 7,734,947 B1
• Filed: 04/17/2007
• Issued: 06/08/2010
• Est. Priority Date: 04/17/2007
• Status: Active Grant

• Current Assignee: NetApp, Inc.
• Original Assignee: NetApp, Inc.
• Inventors: Kownacki, Ronald W., Frangioso, Robert B., Jiang, Tianyu
• Primary Examiner(s): Chu; Gabriel L
• Application Number:
  US11/736,197
• Time in Patent Office: 1,148 Days
• Field of Search: None
• US Class Current: 714/4.1
• CPC Class Codes: G06F 11/1482 by means of middleware or OS functionality
• CPC Class Codes: G06F 11/2025 using centralised failover control functionality
• CPC Class Codes: G06F 2201/815 Virtual

Abstract

A cluster-wide published configuration system and method is provided for assigning host nodes to virtual interfaces (VIFs) in a cluster of data storage systems. A coordinating VIF manager publishes a set of failover rules using a replication database. The coordinating VIF manager monitors VIF hosting status within a cluster. In the event of a VIF failure, the coordinating VIF manager publishes a reassignment of host nodes using a single read-write transaction of a replication database service. Secondary VIF managers in the cluster must perform VIF hosting in accordance to the reassignment. If a secondary VIF manager can not activate an assigned VIF it must publishing new reassignment of VIF hosts via the replication database service.

Claims

1. A method for managing virtual interfaces (VIFs) comprising:

providing a VIF manager instance on each node of a cluster, wherein one of the VIF managers is designated as coordinator;

providing a set of VIF failover rules for all VIFs in the cluster to each VIF manager via a replication database (RDB) service;

using a single read-write (RW) transaction of the RDB service by the coordinator to publish revised VIF hosting assignments pursuant to the VIF failover rules in response to a VIF failure; and

hosting, by each VIF manager of each node affected by the published revised VIF hosting assignments, VIFs pursuant to the published revised VIF hosting assignments without performing additional RW transactions by any VIF manager in the cluster.

2. The method of claim 1 further comprising:

if an affected node cannot host VIFs pursuant to the published revised VIF hosting assignments then,

using the single RW transaction of the RDB service by a VIF manager on the affected node to publish new revised VIF hosting assignments pursuant to the VIF failover rules; and

hosting VIFs by affected nodes pursuant to the published new revised VIF hosting assignments.

3. The method of claim 1 wherein the VIF managers comprise an RDB quorum.

4. The method of claim 3 further comprising:

deactivating, by a particular VIF manager, VIFs hosted by the particular VIF manager when the particular VIF manager drops out of quorum.

5. The method of claim 1 further comprising:

delaying publication of the revised VIF hosting assignments for a time sufficient to ensure that the VIF failure is not a transient failure.

6. The method of claim 1 comprising:

calculating the new VIF hosting assignments by the coordinator based on remaining healthy VIF managers in accordance with the failover rules.

7. The method of claim 1 comprising:

receiving an RDB update notification by each secondary VIF manager instance upon publication by the coordinator of the revised VIF hosting assignments; and

in response to receiving the RDB update notification, scanning by each VIF manager its local database to learn its new hosting responsibilities, the receiving and scanning being performed using read-only (RO) transactions.

8. The method of claim 1, wherein each node comprises a data storage system.

9. A cluster of data storage system nodes, each node comprising:

a replication database service; and

a virtual interface (VIF) manager in communication with the replication database service;

the replication database service including a VIF assignment table and a set of failover rules for each of a plurality of VIFs in the cluster; and

a failover process designating one of the VIF managers in the cluster as coordinator, wherein the coordinator publishes a revised VIF assignment table according to the failover rules in response to a VIF failure, the publication being performed by a single read-write transaction, wherein each VIF manager of each node affected by the published revised VIF hosting assignments hosts VIFs without performing additional RW transactions.

10. The cluster according to claim 9, wherein the revised VIF assignment table is calculated by the coordinator according to the failover rules and the available healthy VIF managers.

11. The cluster according to claim 9, wherein the failover process comprises:

an RDB update notification received by each VIF manager instance upon publication by the coordinator of the revised VIF assignment table; and

in response to receiving the RDB update notification, each VIF manager to scan its local database to learn its new hosting responsibilities, wherein read-only (RO) transactions are used to receive the RDB update notification and scan the local database.

12. The cluster according to claim 11, wherein the failover process comprises:

each VIF manager to perform new VIF hosting responsibilities in response to the revised VIF assignment table.

13. The cluster according to claim 12, wherein in response to a VIF manager being unable to perform its new hosting responsibility, a new revised VIF assignment table is published by the VIF manager using a single read-write transaction of the RDB service.

14. The cluster according to claim 13, wherein the new revised VIF assignment table is calculated by the VIF manager according to the failover rules and the available healthy VIF managers.

15. A system for managing virtual interface (VIF) failover comprising:

a VIF manager instance on each node of a network cluster, wherein one of the VIF managers is designated as coordinator;

a set of VIF failover rules for all VIFs in the cluster to each VIF manager via a replication database (RDB) service;

means for using a single read-write (RW) transaction of the RDB service by the coordinator to publish revised VIF hosting assignments pursuant to the VIF failover rules in response to a VIF failure; and

means for hosting VIFs by affected nodes pursuant to the published revised VIF hosting assignments without performing additional RW transactions by any VIF manager instance in the cluster unless an affected node cannot host VIFs pursuant to the published revised VIF hosting assignments.

16. A computer readable medium containing executable program instructions executed by a processor, comprising:

program instructions that configure a virtual interface (VIF) failover manager instance on each node of a cluster, wherein one of the VIF managers is designated as coordinator;

program instructions that provide a set of VIF failover rules for all VIFs in the cluster to each VIF manager via a replication database (RDB) service;

program instructions that use a single read-write (RW) transaction of the RDB service by the coordinator to publish revised VIF hosting assignments pursuant to the VIF failover rules in response to a VIF failure; and

program instructions that host, by each VIF manager of each node affected by the published revised VIF hosting assignments, VIFs pursuant to the published revised VIF hosting assignments without performing additional RW transactions by any VIF manager instance in the cluster unless an affected node cannot host VIFs pursuant to the published revised VIF hosting assignments.

17. A computer readable medium according to claim 16 further comprising:

program instructions that use a single read-write (RW) transaction of the RDB service by a VIF manager on an affected node to publish new revised VIF hosting assignments pursuant to the VIF failover rules if the affected node cannot host VIFs pursuant to the published revised VIF hosting assignments; and

program instructions that host VIFs by affected nodes pursuant to the published new revised VIF hosting assignments.

18. The computer readable medium according to claim 16 further comprising:

program instructions that deactivate, by a particular VIF manager, VIFs hosted by the particular VIF manager when the particular VIF manager drops out of quorum.

19. The computer readable medium according to claim 16 further comprising:

program instructions that delay publication of the revised VIF hosting assignments for a time sufficient to ensure that the VIF failure is not a transient failure.

20. The computer readable medium according to claim 16 further comprising:

program instructions that calculate the new VIF hosting assignments by the coordinator based on remaining healthy VIF managers in accordance with the failover rules.

21. A method for managing virtual interfaces (VIFs) comprising:

providing a VIF manager instance on a plurality of network elements, wherein one of the VIF managers is designated as coordinator;

providing a set of VIF failover rules for VIFs in the network to the VIF managers via a replication database (RDB) service;

using a single read-write RDB transaction by the coordinator to publish revised VIF hosting assignments pursuant to the VIF failover rules in response to a VIF failure and implement hosting of VIFs according to the revised hosting assignments; and

using read-only RDB transactions, without using any additional read-write RDB transactions, to respond to the revised VIF hosting assignments.

22. The method of claim 21, wherein the plurality of network elements include members of the set consisting of nodes, routers, N-modules, D-modules and switches.

23. A method for managing virtual interface (VIF) failover comprising:

publishing failover rules in a replicated database;

using read-only RDB transactions to respond to the publication of failover rules; and

moving failover enabled data VIFs to an available network port specified in a failover rule in response to a node suffering a failure.

24. A method for managing virtual interfaces (VIFs) comprising:

providing a VIF manager instance on each node of a cluster, wherein one of the VIF managers is designated as coordinator;

providing a set of VIF failover rules for all VIFs in the cluster to each VIF manager via a replication database (RDB) service;

using a first single read-write (RW) transaction of the RDB service by the coordinator to publish revised VIF hosting assignments pursuant to the VIF failover rules in response to a VIF failure and to host VIFs by the affected nodes pursuant to the published VIF hosting assignments;

in response to an affected node being unable to host VIFs pursuant to the published revised VIF hosting assignments, utilizing a second single RW transaction of the RDB service by a VIF manager on the affected node to publish new revised VIF hosting assignments pursuant to the VIF failover rules; and

hosting VIFs by affected nodes pursuant to the published new revised VIF hosting assignments.

25. A method for managing virtual interfaces (VIFs) comprising:

providing a VIF manager instance on each node of a cluster, wherein one of the VIF managers is designated as coordinator, and the VIF managers comprise an RDB quorum;

deactivating, by a particular VIF manager, VIFs hosted by the particular VIF manager when the particular VIF manager drops out of quorum;

providing a set of VIF failover rules for all VIFs in the cluster to each VIF manager via a replication database (RDB) service; and

using a single read-write (RW) transaction of the RDB service by the coordinator to publish revised VIF hosting assignments pursuant to the VIF failover rules in response to a VIF failure and to host VIFs by the affected nodes pursuant to the published VIF hosting assignments.

26. A cluster of data storage system nodes, each node comprising:

a replication database service;

a virtual interface (VIF) manager in communication with the replication database service;

the replication database service including a VIF assignment table and a set of failover rules for each of a plurality of VIFs in the cluster; and

a failover process designating one of the VIF managers in the cluster as coordinator, wherein the coordinator publishes a revised VIF assignment table according to the failover rules in response to a VIF failure using a first single-read write transaction, wherein each VIF manager performs new VIF hosting responsibilities in response to the revised VIF management table, and in response to a VIF manager being unable to perform its new hosting responsibility, a new revised VIF assignment table is published by the VIF manager using a second single read-write transaction of the RDB service.