What's in this release

What's new

Virtualization

• Virtual machine high availability (HA)
  • Automatically reboots a virtual machine when the system detects a virtual machine operating system failure.
  • Allows for configuring the HA failure detection sensitivity for clusters.
• Supports fetching and displaying the usage of virtual machine file system partitions using Arcfra VMTools.
• Supports setting the virtual machine CPU compatibility to SkyLake-noMPX for clusters.

Block storage

• Offers the following enhancements on storage space policies:
  • Supports configuring erasure coding as the redundancy policy for virtual volumes or NFS files in a cluster deployed with tiered configuration. This support only applies to clusters that are not active-active clusters.
  • Displays the overall storage efficiency, space usage, and data reduction savings for clusters.
  • Provides volume pinning for active-active clusters.
  • Supports converting thin provisioning to thick provisioning for virtual volumes online.
• Allows for mounting data disks of up to 256 TB on a single node.

Networking

• Supports configuring MTU for VM networks and the associated virtual NICs.
• Allows for manually specifying the active port on each host when the virtual distributed switch adopts the active-backup bonding mode.

Operations and management

• Supports installing Arcfra Cloud Operating System (ACOS) on independent physical disks, and protects system disk redundancy with hardware RAID 1 configuration.
• Supports modifying the default password of the ACOS operating system when deploying and scaling a cluster.
• Supports displaying the current system time in clusters and hosts on Arcfra Operation Center (AOC).

Improvements

Virtualization

• Optimizes the cross-cluster hot migration for virtual machines to improve the stability of migration tasks and cluster operations during migration.
• Retains the virtual machine UUID unchanged during cross-cluster cold migration or segmented migration.
• Supports uploading the UDF format ISO images to the content library.
• Improves the automatic scheduling logic. After configuring the VM placement group policy that prioritizes placing the virtual machines on different hosts, virtual machines will be evenly placed on the selected nodes.

Block storage

• Improves the storage performance in different scenarios:
  • Optimizes the cache mechanism to improve the cache hit rate for write requests.
  • Optimizes performance for write-heavy workload. Improves the cluster performance in large data import scenarios and reduces the performance impact on other volumes.
  • Optimizes the small block write performance after taking the snapshot.
  • Improves the write performance between availability zones in active-active clusters via compression.
  • Improves the write performance for damaged data areas in post-failure scenarios.
• Increases the upper limit of the reserved cache ratio for volume pinning to 50%.
• Improves the recovery policy elasticity:
  • Speeds up data recovery and rebalancing when using high-end hardware.
  • Improves the recovery success when using less capable hardware.
• In clusters without Boost mode enabled, reduces the I/O interruption time for maintenance scenarios that trigger host maintenance mode to less than 1 second to prevent significant I/O interruptions on the client side.

Networking

Supports editing the VLAN ID of the storage network and the physical ports associated with the virtual distributed switch where the storage network with RDMA disabled resides on AOC.

Operations and management

• Allows for not mounting specified physical disks when deploying or scaling a cluster.

• Optimizes the faulty node replacement process:

  • Achieves data synchronization and node reconstruction by removing the node and adding a node with the same management IP address, storage IP address, and access IP address.
  • Supports role switching and node replacement in case of multiple master node abnormalities.

• Optimizes log collection by using the sosreport tool to gather information on hardware firmware, system performance, memory, and more.

• Proactively takes the faulty disks offline in I/O blocking scenarios, preventing single disk failures from affecting the overall I/O of the node.

• Improves the identification logic of USB devices to ensure that different devices with the same serial number can be differentiated from each other.