What ceph/ceph shipped
Written by FoxPlug from public releases; not affiliated with Ceph. An automatic summary of the public release, pull request and commit data of github.com/ceph/ceph. Ceph did not write it and does not use or endorse FoxPlug. Every line links to the public change it describes.
Get a weekly update like this for your product, free
Week of September 21, 2026
What shipped
- Hardened cephx service ticket verification to reject tickets with modified fields after decryption. Pull request #71231
- Crimson OSD now reads Prometheus metrics port from Ceph config instead of requiring a command-line flag. Pull request #71993
- Added RGW role permission policy API endpoint to mgr/dashboard for managing IAM policies. Pull request #72002
- Centralized cephadm service dependency calculation so deployment and reconciliation use the same logic. Pull request #71655
- Fixed udev rules to skip ceph-rbdnamer on RBD device removal events. Pull request #71467
- Fixed rbd snap purge iterator invalidation when filtering protected snapshots. Pull request #71932
- Fixed write ordering issue in OSD when error injection interacts with scrub-blocked writes. Pull request #71187
- Fixed OSD peering bug where pwlc could retain values larger than
last_update. Pull request #71579 - Reduced repetitive INFO logs from
rbd_supportschedule handlers by moving periodic messages to DEBUG level. Pull request #70781 - Bumped NFS-Ganesha version to 15 to enable BYOK and delegation support. Pull request #71804
Why it matters
This week includes security hardening for cephx authentication, infrastructure improvements for Crimson and cephadm, and several critical bug fixes affecting data integrity in OSD peering and RBD operations. The RGW IAM policy API addition expands dashboard management capabilities for multi-tenant deployments.
Changelog entry
- auth/cephx: Harden service ticket verification to reject tickets with modified fields Pull request #71231
- crimson/osd: Read Prometheus metrics port from Ceph config instead of command-line flag Pull request #71993
- mgr/dashboard: Add RGW role permission policy API endpoint Pull request #72002
- mgr/cephadm: Centralize service dependency calculation Pull request #71655
- udev: Skip ceph-rbdnamer on RBD device removal events Pull request #71467
- rbd: Fix iterator invalidation in snap purge when filtering protected snapshots Pull request #71932
- osd: Fix write ordering issue with error injection and scrub-blocked writes Pull request #71187
- osd: Fix pwlc retaining values larger than
last_updateduring peering Pull request #71579 - mgr/rbd_support: Move periodic schedule-loading messages to DEBUG level Pull request #70781
- container: Bump NFS-Ganesha version to 15 for BYOK and delegation support Pull request #71804
This week: cephx security hardening, Crimson Prometheus config support, RGW IAM policy API, centralized cephadm dependencies, and critical OSD peering fixes.
Ceph shipped significant improvements this week: hardened cephx service ticket verification for security, Crimson OSD now reads Prometheus port from config, dashboard gained RGW role policy management, and critical fixes for OSD write ordering and peering bugs. Service dependency logic is now centralized in cephadm for consistency.
Week of September 14, 2026
What shipped
- SimpleRADOSStriper now correctly captures asynchronous I/O errors by explicitly calling get_return_value() after wait_for_complete() instead of incorrectly assuming the wait function returns the operation status. Pull request #69195
- SimpleRADOSStriper fixed a resource leak in the
shrink_allocerror path where un-awaited librados::AioCompletion objects were leaked when an AIO wait failed. Pull request #69197 - SimpleRADOSStriper corrected allocation growth calculation in set_metadata() to apply the base-2 rounding mask to the requested
new_sizeinstead of the old size. Pull request #69194 - OSD
rep_repair_primary_objectnow handles balanced or localized reads sent to replicas that fail by having the replica fail the operation rather than asserting. Pull request #70909 - RocksDB submodule bumped from v7.10.2 to v9.11.2 with cmake requiring RocksDB 9.10 or later. Pull request #70807
- Crimson OSD now prevents write omap operations on pools that do not support omap, failing with -EOPNOTSUPP. Pull request #71874
- OSDMap::calc_pg_maps fixed upmap balancing to accept changes that do not decrease stddev and to properly value the deviation of small OSDs. Pull request #65191
- OSDC added minimum allowable split size for replica IOs to prevent issues with the
prepare_single_oppath. Pull request #69290 - MGR get_daemon_health_metrics() now properly takes DaemonState::lock when iterating over daemon health metrics. Pull request #71114
- DaemonServer::shutdown() fixed race conditions and resource leaks including ensuring the SafeTimer thread is properly joined before destruction. Pull request #71234
Why it matters
This week brings fixes for I/O error handling in RADOS striping, OSD replication and balancing, resource leaks in daemon shutdown, and RocksDB dependency updates. Several bugs affecting data consistency and system stability were addressed across core storage and management components.
Changelog entry
- SimpleRADOSStriper: retrieve AIO return values with get_return_value() Pull request #69195
- SimpleRADOSStriper: ensure all AIO completions are awaited in
shrink_allocerror path Pull request #69197 - SimpleRADOSStriper: fix incorrect variable usage in allocation growth calculation Pull request #69194
- SimpleRADOSStriper: fix missing thread synchronization headers Pull request #69196
- OSD:
rep_repair_primary_objectnow copes with balanced/localized reads Pull request #70909 - Crimson OSD: do not allow executing write omap operations on unsupported pools Pull request #71874
- OSDMap: fix balancing for upmap corner cases with stddev stalling Pull request #65191
- OSDC: add minimum allowable split size for replica IOs Pull request #69290
- RocksDB: bump version from v7.10.2 to v9.11.2 Pull request #70807
- MGR: take DaemonState lock in get_daemon_health_metrics() Pull request #71114
- DaemonServer: fix race conditions and resource leaks in shutdown() Pull request #71234
- Build: link cephfs-tool against Boost::program_options Pull request #71712
- MGR/crash: fix prune_detail() not actually pruning the list Pull request #71181
- Tracing: fix OSD segmentation faults by adding
bluestore_tpto install list Pull request #69341
Week 40: SimpleRADOSStriper fixes for AIO error handling and resource leaks, OSD replication improvements, upmap balancing corrections, and RocksDB v9.11.2 integration.
This week's Ceph updates focus on reliability and resource management. Key fixes include SimpleRADOSStriper properly capturing asynchronous I/O errors and preventing resource leaks, OSD replication handling balanced reads on replicas correctly, and upmap balancing improvements. Infrastructure changes include RocksDB dependency bump to v9.11.2 and daemon shutdown race condition fixes.