Server management

The mmvdisk command considers a node to be a server if the node is a member of an mmvdisk node class, or if the node has a non-zero nsdRAIDTracks configuration setting.

Before a recovery group can be created, the servers in a node class should be verified to have the same memory and the same expected disk topology, and they must be configured to enable IBM Storage Scale RAID.

The memory configuration of a server node class can be checked using the mmvdisk server list --config command:


mmvdisk server list --node-class ESS01 --config

 node
number  server                            active   memory    pagepool  nsdRAIDTracks
------  --------------------------------  -------  --------  --------  -------------
     1  ess01io1.gpfs.net                 yes       123 GiB  1024 MiB              0
     2  ess01io2.gpfs.net                 yes       123 GiB  1024 MiB              0

The disk topology of a server node class can be checked using the mmvdisk server list --disk-topology command:


mmvdisk server list --node-class ESS01 --disk-topology

 node                                       needs    matching
number  server                            attention   metric   disk topology
------  --------------------------------  ---------  --------  -------------
     1  ess01io1.gpfs.net                 no          100/100  ESS GL6
     2  ess01io2.gpfs.net                 no          100/100  ESS GL6

Note: Recognition of server disk topologies depends on the appropriate IBM Storage Scale System 3200, IBM Spectrum® Scale Erasure Code Edition, or IBM Elastic Storage® System 3000 support RPMs being installed on all server nodes and on any node where the mmvdisk command is run.

The node class is configured using the mmvdisk server configure command:


mmvdisk server configure --node-class ESS01 --recycle one
mmvdisk: Checking resources for specified nodes.
mmvdisk: Node 'ess01io1.gpfs.net' has a paired recovery group disk topology.
mmvdisk: Node 'ess01io2.gpfs.net' has a paired recovery group disk topology.
mmvdisk: Node class 'ESS01' has a paired recovery group disk topology.
mmvdisk: Setting configuration for node class 'ESS01'.
mmvdisk: Node class 'ESS01' is now configured to be recovery group servers.
mmvdisk: Restarting GPFS daemon on node 'ess01io1.gpfs.net'.
mmvdisk: Restarting GPFS daemon on node 'ess01io2.gpfs.net'.

The mmvdisk server configure command verifies that all node class members have the same general type of server disk topology: either a paired recovery group (Elastic Storage Server) disk topology, or a scale-out recovery group (IBM Storage Scale Erasure Code Edition) disk topology, or a shared recovery group (IBM Elastic Storage System 3000) disk topology. Different IBM Storage Scale configuration settings are applied depending on which type is found.

Important differences between the two configurations are that the paired recovery group pagepool and nsdRAIDTracks settings depend on the amount of real memory, but the shared and scale-out recovery group pagepool defaults to 20% of real memory and nsdRAIDTracks is always 131072 (128K). Regardless of recovery group type, the minimum pagepool setting for a server is 8 GiB.

Table 1. Default paired recovery group pagepool and nsdRAIDTracks settings
Real memory	pagepool	nsdRAIDTracks
Between 32 GiB and 127 GiB	60% of real memory	131072 (128K)
Between 127 GiB and 256 GiB	70% of real memory	262144 (256K)
Greater than 256 GiB	70% of real memory	327680 (320K)

The default pagepool setting for an mmvdisk node class can be changed with the --pagepool option when using the mmvdisk server configure command. There is not a similar option for the nsdRAIDTracks setting.

The mmvdisk command uses the mmvdisk node class to maintain the IBM Storage Scale RAID configuration settings for the recovery group servers. The mmvdisk command never uses cluster-wide settings for recovery group servers, since different recovery groups can use different settings. If recovery group server settings are ever made independently of mmvdisk, the same practice of using the mmvdisk node class should be followed.

When a server is added to a scale-out recovery group, the server must first be temporarily configured independently of the recovery group's node class. This is accomplished by using the -N Node option to the mmvdisk server configure command and restarting the daemon on the node. When the node is added to the scale-out recovery group, it becomes a member of the node class and the temporary configuration is removed. For more information on adding a server to a scale-out recovery group, see Recovery group management.