8.1 What’s New and Changed for Clustered Storage

8.1.1 NSS Pools (Storage Manager)

Novell Cluster Services supports NSS pools. In a mixed-mode cluster, existing NSS pool cluster resources can be cluster migrated to any node in the cluster.

NSS pools are managed by different storage managers on each platform, as shown in Table 8-1. In a mixed-mode cluster with older operating systems, the OES 11 (or later) nodes are not able to recognize storage management actions performed on a pool on the alternate management system.

Table 8-1 Storage Managers for NSS Pools

Platform

Storage Manager

OES 11 and later

Novell Linux Volume Manager (NLVM)

OES 2 SP3

Enterprise Volume Management System (EVMS)

NetWare 6.5 SP8

Media Manager

In NetWare 6.5 SP8 and OES 2 SP3, NSS management tools recognize only DOS devices that are up to 2 TB in size. In OES 11 and later, the NSS management tools can initialize a device to use the DOS partition table scheme or GUID partition table (GPT) scheme. DOS supports device sizes up to 2 TB. GPT supports device sizes of any size, which allows you to create NSS pools up to 8 TB (terabytes) in size on a single device. In a mixed-mode cluster, the older platforms do not recognize devices that are partitioned with the GPT partitioning scheme.

WARNING:In a mixed-mode cluster, do not perform any storage management actions (create, modify, or delete) for NSS pools while OES 11 (or later) nodes are active in the cluster. Any such action can result in data corruption.

You must not perform storage management actions on pools while the cluster is in mixed mode. Wait to create new NSS pools until after all of the nodes in the cluster are upgraded. If it is necessary to create, modify, or delete a pool as part of a documented cluster resource upgrade process, you must shut down the OES 11x nodes, create the NSS pool cluster resource on the older nodes, then restart the upgraded nodes. This allows the upgraded nodes to see the new configuration. This is critical so that pools are able to cluster migrate across platforms, and to prevent possible corruption.

For information about creating and managing NSS pool cluster resources on OES 11 or later, see Section 13.0, Configuring and Managing Cluster Resources for Shared NSS Pools and Volumes.

8.1.2 Behavior of an NSS Pool Resource with Media Version 44.03 and Above in Mixed Node Cluster

A mixed node cluster environment can contain OES 11 SP2, OES 2015, and OES 2015 SP1 nodes. The AD media upgraded NSS32 and NSS64 pools cannot be loaded in OES 11 SP2 nodes.

When a resource comes online, it tries to load on a node based on the preferred node assignment (OES 2015 or later nodes), and the older OES cluster nodes are skipped from the preferred node list. If OES 2015 or later nodes are not available in the cluster, the resources are moved to “unassigned” state.

However, if the upgraded (OES 2015 or later) node comes up or any new OES (OES 2015 or later) node is added to the cluster, the resource will automatically loads on the OES 2015 or later node.

IMPORTANT:

  • All nodes in the cluster must be patched with the latest OES patches. Otherwise, the AD media resource goes to comatose state on nodes earlier than OES 2015.

  • All nodes in the cluster must belong to OES 11 SP2 or later. If any OES node in the cluster is older than OES 11 SP2, this feature does not work.

In NSS32 and NSS64 pool type, the pool status can be determined by the media version. Table 8-2 provides information about the NSS32 and NSS64 pool type media version.

Table 8-2 NSS Pool Media Version Details

Pool Media Version

Description

44.03

  • 44: An NSS32 pool that supports AD.

  • 03: All volumes are upgraded to support hard links.

51.00

  • 51: An NSS64 pool that supports AD by default.

  • 00: All volumes are upgraded to support hard links.

45.00

  • 45: An NSS32 pool that supports Trustee Index.

  • 00: Some volumes are not upgraded to support Trustee Index.

52.00

  • 52: An NSS64 pool that supports Trustee Index.

  • 00: Some volumes are not upgraded to support Trustee Index.

45.01

  • 45: An NSS32 pool that supports Trustee Index.

  • 01: All volumes are upgraded to support Trustee Index.

52.01

  • 52: An NSS64 pool that supports Trustee Index.

  • 01: All volumes are upgraded to support Trustee Index.

Similarly, the volume status can also be determined by the media version. Table 8-3 provides information about the NSS volume media version.

Table 8-3 NSS Volume Media Version Details

Volume Media Version

Description

38.05

  • 38: An NSS volume that supports Hard Link.

  • 05: An NSS volume is upgraded to support hard links.

41.00

  • 41: An NSS volume that supports Trustee Index.

  • 00: An NSS volume is not upgraded to support Trustee Index.

41.01

  • 41: An NSS volume that supports Trustee Index.

  • 01: An NSS volume is upgraded to support Trustee Index.

8.1.3 Linux POSIX File Systems on an EVMS CSM (Compatibility Only)

Novell Cluster Services for OES 11 and later provides compatibility support for Linux POSIX file systems that were created by using the Cluster Segment Manager (CSM) in EVMS (Enterprise Volume Management System).

For OES 2 SP3, Novell Cluster Services provided a Cluster Segment Manager (CSM) for EVMS to support cluster resources for Linux POSIX file systems. EVMS was deprecated in the SUSE Linux Enterprise Server (SLES) 11 release, and is no longer available. Accordingly, Novell Cluster Services provides compatibility support for these cluster resources. It has been modified in the following ways to allow you to move clustered Linux POSIX file systems from OES 2 SP3 to OES 11 and later:

  • Recognizes and supports the EVMS-based file system structure.

  • Provides a way for managing a resource in the load, unload, and monitor scripts that works with the EVMS device and segment structure.

  • Changes the device path from /dev/evms/ to /dev/mapper/, which is the same location where LVM devices are stored.

Before any of the new OES 11x nodes join an OES2 SP3 cluster, you must offline all resources that use Linux POSIX volumes. Modify their load, unload, and monitor scripts to use the new commands for compatibility mode. Allow some OES 11x nodes to join the OES 2 SP3 cluster, then modify each resources’ preferred nodes list to use only the OES 11x nodes in the cluster. Ensure that you disable Resource Follows Master setting for the resource until the conversion is finalized. After a resource’s scripts and preferred nodes list have been modified, you can bring the resource online only on OES 11x nodes. As you add other OES 11x nodes, you can add those to the preferred nodes lists for failover.

After a resource has been moved from an OES 2 SP3 node to an OES 11x node, it is not supported to move it back to an OES 2 SP3 node. The resource should fail over only to other OES 11x nodes in a mixed cluster.

In the NSS management tools, the CSM layer is reported in the Details page for the shared device. This allows you to recognize the difference between the Linux POSIX resources that were created with EVMS and the new Linux LVM volume group resources that are created on OES 11x.

For information about managing CSM cluster resources on OES 11 and later servers, see Section 15.0, Upgrading and Managing Cluster Resources for Linux POSIX Volumes with CSM Containers.

8.1.4 Linux POSIX File Systems on an LVM Volume Group

Beginning in OES 11 or later, Novell Cluster Services supports Linux POSIX file systems on Linux Volume Manager (LVM) volume groups, including Btrfs, Ext2, Ext3, ReiserFS, and XFS. For information about the file systems, see the Overview of Linux POSIX File Systems in the OES 2015 SP1: Linux POSIX Volume Administration Guide.

IMPORTANT:You should not create new Linux POSIX cluster resources while the cluster is in mixed mode, except as part of a documented conversion procedure.

The LVM-based cluster resources can be used on OES 11 or later nodes in a mixed-mode cluster. The resource must not be moved to a node running the older platform. Ensure that you set up your preferred nodes for failover so that the LVM volume group cluster resource fails over only to the nodes running the new platform in a mixed-mode cluster.

Beginning with OES 11 SP1 or later, NSSMU and Novell Linux Volume Manager (NLVM) commands can be used to create cluster enabled LVM volume group cluster resources. See Section 14.0, Configuring and Managing Cluster Resources for Shared LVM Volume Groups.

Linux LVM management tools are also available in the YaST Expert Partitioner and in native Linux LVM command line commands. For information about using Linux commands to create a shared LVM volume group and cluster enabling it for Novell Cluster Services, see Section 14.4.1, Creating a Shared LVM Volume with LVM Commands.