Summary of changes

• Asynchronous notifications to detect changes on a cloud object store instead of polling. For more information, see Asynchronous notifications and Amazon SQS integration.
• Configuration AFM to cloud object storage fileset by using proxy endpoints. For more information, see Creating a fileset by using proxy endpoints.
• AFM to cloud object storage fileset with S3 Glacier storage classes support for all fileset modes. For more information, see Configuring an AFM to cloud object storage fileset with S3 Glacier storage classes.
• Support custom certificates for AFM cache filesets with the cloud object storage. For more information, see Configuring AFM filesets with custom SSL certificates.
• Removal of the afmEnableADR parameter. To enable AFM DR, the afmEnableADR parameter is enabled by default. If this parameter is disabled, then AFM DR relationship will not work.

• On Microsoft Azure, automated deployment and configuration of CES or protocol nodes.
• Extended support for RHEL 9.6.
• Cloudkit binary packaging and its dependencies as an RPM package.

Changed commands

• mmaddcallback
• mmadddisk
• mmafmcosconfig
• mmcallhome
• mmces
• mmchattr
• mmchconfig
• mmchfileset
• mmchfs
• mmchlicense
• mmchode
• mmcrfs
• mmfsck
• mmlsattr
• mmlscluster
• mmlsfs
• mmlspool
• mmperfmon
• mmpmon
• mmqos
• mms3
• scalectl filesystem
• scalectl node

Starting with 6.0.0.1, support for the decentralized deployment of data acceleration tier (DAT) in IBM Storage Scale System 6000. For more information, see Decentralized DAT deployment in the IBM Storage Scale System 6000: Hardware Planning and Installation Guide.

• Enhanced online file system checks through improvements to the mmfsckx command. The mmfsckx command can now detect and report inconsistencies in a directory block structure. This feature only detects and reports the directory block inconsistency but does not repair it. For more information, see mmfsckx command.
• Starting with IBM Storage Scale 6.0.0, only for newly created clusters, the default value of prefetchLargeBlockThreshold changes to 4194304. This change causes prefetching to speed up more gradually for file systems which block sizes are equal to or greater than 4 MiB. Typically, this more gradual increase in prefetching speed contributes to read amplification effects. The prefetchLargeBlockThreshold option of the mmchconfig command can be changed without restarting the IBM Storage Scale mmfsd daemon through the mmchconfig -i option. When the prefetchLargeBlockThreshold option is set, it defines the minimum block size of a file system, for which a new more gradual (less abrupt) prefetching mode is enabled. For clusters created before version 6.0.0, the default value of prefetchLargeBlockThreshold is 33554432.
• New parameter for the mmchconfig command to set the TCP communications mode

  Specifies the TCP connections mode: multi-rail over TCP (MROT) or multiple connections over TCP (MCOT). For more information, see mmchconfig command in the IBM Storage Scale: Command and Programming Reference Guide.

• Expelled nodes:
  • When a GPFS node is expelled by the cluster manager, the mmfs.log message now provides clearer details about the reason for the expel decision.
  • A new postExpel event is introduced. If a GPFS node leaves the cluster unexpectedly (for example, due to lost leases, mmexpelnode invocations, or inter-node RPC failures), this event allows users to optionally trigger customized logging and post-expel actions through a callback.
  • The mmhealth component also reports such expels through its own postExpel callback.
• Bash completion added for mmhealth, mmcallhome, and mmperfmon.

• Added the scalectl node batchAdd command and the /scalemgmt/v3/nodes:batchAdd: POST endpoint for native REST API-enabled clusters to add multiple nodes with JSON or node descriptor file input. For more information, see scalectl node command and /scalemgmt/v3/nodes:batchAdd: POST.
• Added the scalectl filesystem quota command and the /scalemgmt/v3/filesystems/{filesystem}/quotas endpoints for file system quota management. For more information, see scalectl filesystem command and Quota management endpoints.
• Added support for ARM architecture.

For more information, see IBM Storage Scale management API endpoints.

• Modernization of GUI towards IBM’s Carbon standard with enhancements to node and events page.
• Support for ARM.
• Support for nodes booted in FIPS mode.
• GUI support for JBOF and DAT configurations.
• Improvements to support the creation of many filesets through the REST API.

For more information, see Introduction to IBM Storage Scale GUI.

• Perfmon performance is improved for large-scale queries involving many filesets.
• Clock alignment added for sensors to synchronize measurements across nodes.
• REST API now supports measurement queries with aggregation and group-by.
• Prometheus exporter now supports query filtering to reduce data load.

• Updated Samba on IBM Storage Scale to the upstream Samba 4.22 version.
• Implemented SMB file listing optimization by using asynchronous threads.
• If slowness is observed during listing a directory with large number of files or directories (>= 50000), optimization can be enabled by changing smbd async dosmode at global or share levels to "yes" by using the mmsmb command.
• New SMB configuration option, vfs mkdir use tmp name is set to no by default to synchronize with older versions and avoid errors in notifications.
• The mmuserauth command in IBM Storage Scale is tightly coupled with SMB 4.22 version when you use the --enable-nfs-kerberos option. You cannot use this command with older SMB versions.

• AFM replication errors now detected and reported via mmhealth.
• Node expel monitoring with expel details added.
• TIPs added for filesystem -n mismatch and quorum node redundancy.
• New default threshold for network errors.
• Added S3 webhook connection health monitoring and warnings when exceeding S3 account or bucket limits.
• mmpmon socket and proxy failures now detected and reported.
• mmhealth node show --unhealthy now hides headers of healthy nodes.

• The QoS feature to rebalance I/O limits amongst nodes has been renamed to Dynamic QoS Rebalancing (DQR). DQR was previously known as Mount Dynamic I/O (MDIO) in IBM Storage Scale versions prior to 6.0.0. Existing MDIO configuration setting mdio-enabled for user classes and the throttle for mdio-all-sharing to enable QoS class sharing remain valid and compatible with DQR. However, it is recommended to use the new DQR attributes for any future changes. If the cluster's minReleaseLevel is 6.0.0 or higher, attributes and class names will display the dqr prefix; otherwise, they will display the legacy mdio prefix.
• Filesystem-level QoS feature is available as a technology preview in IBM Storage Scale 6.0.0. For instructions on how to enable and run the Filesystem-level QoS feature, sign up for the technology preview by contacting scale@us.ibm.com.