Ceph’s v19.2.0 Squid marks the 19th stable release, designed to continue Ceph’s leadership in scalable, distributed storage. Squid brings a blend of performance tweaks, new management tools, and under-the-hood fixes aimed at improving the overall Ceph experience for storage administrators and developers alike.

This release builds on Ceph’s last release, Reef, and introduces optimizations for BlueStore, Ceph’s underlying storage engine. In addition, it improves CephFS (Ceph’s file system), making snapshot management, resource management, and client access simpler and more powerful.

However, Ceph also warns users, especially those relying on iSCSI, to be cautious while upgrading from previous versions due to a known bug during the transition from 19.1.1 to 19.2.0. For more information, you can refer to the official release notes or reach out to 42on for consultancy services.

Major Improvements in Ceph v19.2.0

This release features key upgrades across various components. Some of the most notable changes include:

• Performance enhancements: BlueStore has been optimized for better handling of snapshot-heavy workloads.
• RADOS changes: New scrub scheduling behavior and an improved CRUSH rule type for more flexible EC (Erasure Coding) configurations.
• CephFS: Enhancements to managing snapshots, subvolumes, and metadata operations.
• RBD (RADOS Block Device): Performance improvements in QEMU live disk synchronization, particularly for backups.
• RGW (RADOS Gateway): New AWS-compatible IAM APIs and multi-site replication fixes for encrypted S3 objects.

Performance enhancements

For storage workloads that involve a large number of snapshots, BlueStore has received substantial improvements. LZ4 compression in RocksDB is now enabled by default, boosting performance and reducing space consumption for high-speed devices.

New tools for iSCSI and CephFS

CephFS, the distributed file system layer of Ceph, has received new features allowing admins to better manage snapshots and clones. This allows for more fine-grained control of file system resources and makes mounting volumes simpler with new helper commands.

However, a known bug in iSCSI during the upgrade process should be considered, especially for users upgrading from 19.1.1 to 19.2.0. Check Ceph’s tracker issue 68215 for more on the potential pitfalls and solutions.

Upgrading from previous versions

When upgrading from Reef or Quincy, it is important to follow the recommended steps to ensure the smoothest transition. The Ceph community advises first stabilizing your cluster and closely monitoring the upgrade process, especially if using non-Cephadm clusters.

• Use the command

ceph orch upgrade start --image quay.io/ceph/ceph:v19.2.0

for Cephadm-managed clusters.

• The upgrade is fully automated for Cephadm deployments, and users are provided real-time progress tracking with ceph -s or ceph orch upgrade status.

For clusters not using Cephadm, the process involves manually upgrading key daemons such as monitors, OSDs, and managers. If you need assistance with this, feel free to call us for support.

RADOS improvements in Squid

RADOS, the core storage system in Ceph, has seen a host of updates in this release. Key changes include:

• New scrubscheduling optimizations to enhance performance in large clusters.
• Introduction of a new CRUSH rule type for better Erasure Coding configurations, enabling more efficient data distribution across pools.
• Improved handling of snapshot-heavy workloads, boosting the overall performance of RADOS operations.

CephFS upgrades

Ceph’s file system, CephFS, receives several exciting features, making it easier for administrators to manage distributed file systems and applications:

• Support for crash-consistent snapshots: This feature is useful for distributed applications that require consistency across multiple clients.
• New commands to pause I/O and metadata mutations, supporting better control of application states during backups or maintenance.

In addition, improvements in the CephFS subvolume system now make it possible to pause specific trees, pause operations, and support new management functionalities that were previously unavailable.

RBD enhancements

The RADOS Block Device (RBD) module has introduced several performance improvements:

• Diff-iterate can now execute locally, which dramatically speeds up QEMU live disk synchronization for large-scale backups.
• Multiplexing of image mappings, reducing TCP session overhead and improving overall system resource management.

These enhancements position RBD to be more efficient, particularly in virtualized environments that require fast and reliable disk synchronization.

RGW new features

The RADOS Gateway (RGW) component, which enables Ceph to serve as a scalable object storage service, now includes several new AWS-compatible IAM APIs:

• Self-service management of users, groups, roles, and policies for improved access control.
• Better replication handling for S3 multipart uploads, ensuring encrypted objects are replicated accurately in multi-site environments.

These updates are crucial for users running Ceph in enterprise environments where interoperability with AWS services and secure multi-site operations are priorities.

Dashboard improvements

The Ceph Dashboard has undergone a UI/UX overhaul, making navigation more intuitive for system administrators. Key enhancements include:

• Improved layout for managing CephFS snapshots and RGW buckets.
• New helpers to simplify mounting CephFS volumes, making everyday management tasks easier for users.

Crimson and Seastore preview

A key highlight of the Squid release is the tech preview of Crimson/Seastore, which promises to support RBD workloads on replicated pools. This preview is intended for developers interested in testing the cutting-edge architecture of Ceph that focuses on low-latency storage operations.

iSCSI bug notice and workarounds

As highlighted in the release notes, there is a critical bug impacting iSCSI users who plan to upgrade from 19.1.1 to 19.2.0. Before upgrading, it is essential to consult tracker issue 68215 for detailed instructions on how to avoid issues that might disrupt storage operations.

Telemetry and monitoring enhancements

The telemetry channel now captures more detailed insights, including the usage of newer features like Crimson. This enhancement helps the Ceph team understand user adoption trends better, driving future developments.

Upgrade process: Key steps

For users running Quincy or Reef, the upgrade to Squid is straightforward but requires careful monitoring. Whether using Cephadm or manually upgrading, it is important to follow the step-by-step upgrade instructions to ensure a smooth transition.

What users should know before upgrading

• Ensure that the autoscaler is disabled before starting the upgrade.
• Monitor the upgrade’s progress via the cephversions command to check the versions running on each daemon.
• For non-Cephadm users, monitor the upgrade of critical components like monitors, OSDs, and managers.

Each new Ceph release may result in older versions losing support. It’s crucial to verify which versions are no longer supported with each update and to avoid falling too far behind, as this can introduce security and stability risks. 42on can assist you in staying up to date, even if you’re using an older version. We provide upgrade support and ensure that your cluster continues to operate smoothly, regardless of the version you are running.

Key changes for developers

Developers will find several important updates in Squid, including deprecated features like cls_cxx_gather and the move towards Opentracing for better tracing support.

Future outlook for Ceph development

The Squid release is not just an update but a foundation for future innovations, particularly with the Crimson project. As Ceph continues to evolve, expect further optimizations in performance, usability, and features designed for modern workloads.