# v10.2.7



* Backport #17786: jewel: multisite: assertion in RGWRados::wakeup_data_sync_shards
* Backport #17970: jewel: osd/ECBackend.cc: 201: FAILED assert(res.errors.empty())
* Backport #18116: jewel: common osd: Improve scrub analysis, list-inconsistent-obj output and osd-scrub-repair test
* Backport #18502: jewel: Don't lose deep-scrub information
* Backport #18657: jewel: Fix OSD network address in OSD heartbeat_check log message
* Backport #18714: jewel: rgw: multipart uploads copy part support
* Backport #18781: jewel: radosgw swift: error messages: spurious newline after http body causes weird errors.
* Backport #18794: jewel: Prefer backfill for inactive PGs over recovery
* Backport #18811: jewel: librgw: RGWLibFS::setattr fails on directories
* Backport #18901: jewel: librgw:  path segments neglect to ref parents
* Backport #18983: jewel: rgw: sending Content-Length in 204 and 304 responses should be controllable
* Backport #19003: jewel: civetweb defaults to libssl.so and libcrypto.so when versions not passed
* Backport #19038: jewel: [rbd-mirror] deleting a snapshot during sync can result in read errors
* Tasks #19055: jewel v10.2.7
* Backport #19098: RGW/openssl fix for autoconf logic problem in Ubuntu Xenial
* Backport #19115: jewel: rgw_file: ensure valid_s3_object_name for directories
* Bug #19128: rbd import needs to sanity check auto-generated image name
* Bug #19138: "Ceph -w" may have overlapped output when there are too many PG status
* Backport #19143: jewel: rgw_file:  FHCache residence check should be exhaustive
* Backport #19148: jewel: rgw daemon's DUMPABLE flag is cleared by setuid preventing coredumps
* Backport #19152: jewel: rgw_file:  restore (corrected) fix for dir "partial match" (return of FLAG_EXACT_MATCH)
* Backport #19153: jewel: rgw_file:  fix recycling of invalid mkdir handles
* Backport #19159: jewel: multisite: RGWMetaSyncShardControlCR gives up on EIO
* Backport #19161: jewel: rgw_file: fix marker computation
* Backport #19165: jewel: rgw_file:  "exact match" invalid for directories, in RGWLibFS::stat_leaf()
* Backport #19167: jewel: rgw_file:  RGWReaddir (and cognate ListBuckets request) don't enumerate multi-segment directories
* Backport #19169: jewel: rgw_file:  allow setattr on placeholder (common_prefix) directories
* Backport #19176: jewel: swift API: cannot disable object versioning with empty X-Versions-Location
* Bug #19277: Legacy buckets cannot enable versioning
* Backport #19341: jewel: 'period update' does not remove short_zone_ids of deleted zones
* Bug #19388: mount.ceph does not accept -s option
* Backport #19418: jewel: rgw_file:  RGWFileHandle dtor must also cond-unlink from FHCache
* Backport #19423: jewel: librbd/ExclusiveLock.cc: 457: FAILED assert(m_state == STATE_ACQUIRING || m_state == STATE_POST_ACQUIRING || m_state == STATE_WAITING_FOR_PEER)
* Bug #19486: Rebalancing can propagate corrupt copy of replicated object 
* Bug #19823: ceph-disk prepare/activate does not work properly using setuser/setgroup (filestore)
* Bug #20054: librbd memory overhead when used with KVM
* Bug #20178: df reports negative disk "used" value when quota exceed
* Bug #20189: fs
* Bug #20217: cephfs can be mounted even when keyring is modified