v12.2.12 Luminous

This is the twelfth bug fix release of the Luminous v12.2.x long term stablerelease series. We recommend that all users upgrade to this release.

Notable Changes

• In 12.2.11 and earlier releases, keyring caps were not checked for validity,so the caps string could be anything. As of 12.2.12, caps strings arevalidated and providing a keyring with an invalid caps string to, e.g.,ceph auth add will result in an error.

Changelog

• auth: ceph auth add does not sanity-check caps (issue#22525, pr#24906, Jing Li, Nathan Cutler, Sage Weil)

• build/ops: Allow multi instances of “make tests” on the same machine (issue#36737, pr#26186, Kefu Chai)

• build/ops: rpm: require ceph-base instead of ceph-common (issue#37620, pr#25810, Sébastien Han)

• ceph-volume: add –all flag to simple activate (pr#26656, Jan Fajerski)

• ceph-volume: look for rotational data in lsblk (pr#26989, Andrew Schoen)

• ceph-volume: replace testinfra command with py.test (pr#26824, Alfredo Deza)

• ceph-volume: revert partition as disk (issue#37506, pr#26295, Jan Fajerski)

• ceph-volume: simple scan will now scan all running ceph-disk OSDs (pr#26857, Andrew Schoen)

• ceph-volume: use our own testinfra suite for functional testing (pr#26703, Andrew Schoen)

• CLI: ability to change file ownership (issue#38370, pr#26758, Sébastien Han)

• client: session flush does not cause cap release message flush (issue#38009, pr#26271, Patrick Donnelly)

• common: ceph_timer: stop timer’s thread when it is suspended (issue#37766, pr#26579, Peng Wang)

• common: fix for broken rbdmap parameter parsing (issue#36327, pr#26000, Marc Schoechlin)

• core: Objecter::calc_op_budget: Fix invalid access to extent union member (issue#37932, pr#26064, Simon Ruggier)

• core: os/filestore: ceph_abort() on fsync(2) or fdatasync(2) failure (issue#38258, pr#26871, Sage Weil)

• crypto: don’t use PK11_ImportSymKey() in FIPS mode (issue#38843, pr#27104, Radoslaw Zarzynski)

• Fix recovery and backfill priority handling (issue#27985, issue#38041, pr#26793, Sage Weil, xie xingguo, David Zafman)

• journal: max journal order is incorrectly set at 64 (issue#37541, pr#25955, Mykola Golub)

• librgw: export multitenancy support (issue#37928, pr#25986, Tao Chen)

• mds: broadcast quota message to client when disable quota (issue#38054, pr#26293, Junhui Tang)

• mds: fix potential re-evaluate stray dentry in _unlink_local_finish (issue#38263, pr#26473, Zhi Zhang)

• mds: handle fragment notify race (issue#36035, pr#25990, “Yan, Zheng”)

• mds: handle state change race (issue#37594, pr#26005, “Yan, Zheng”)

• mds: log evicted clients to clog/dbg (issue#37639, pr#25858, Patrick Donnelly)

• mds: log new client sessions with various metadata (issue#37678, pr#26257, Patrick Donnelly)

• mds: message invalid access (issue#38488, pr#26661, Patrick Donnelly)

• MDSMonitor: do not assign standby-replay when degraded (issue#36384, pr#26642, Patrick Donnelly)

• MDSMonitor: missing osdmon writeable check (issue#37929, pr#26065, Patrick Donnelly)

• mds: optimize revoking stale caps (issue#38043, pr#26278, “Yan, Zheng”)

• mds: stopping MDS with a large cache (40+GB) causes it to miss heartbeats (issue#37723, issue#38022, pr#26232, Patrick Donnelly)

• mds: trim cache after journal flush (issue#38010, pr#26215, Patrick Donnelly)

• mds: wait for client to release shared cap when re-acquiring xlock (issue#38491, pr#27024, “Yan, Zheng”)

• mds: wait shorter intervals if beacon not sent (issue#36367, pr#25979, Patrick Donnelly)

• mgr: “balancer execute” only requires read permissions (issue#25345, pr#25768, John Spray)

• mgr/balancer: restrict automatic balancing to specific weekdays (pr#26501, xie xingguo)

• mgr/BaseMgrModule: drop GIL for ceph_send_command (issue#38537, pr#26830, Sage Weil)

• mgr/DaemonServer: log pgmap usage to cluster log (issue#37886, pr#26207, Neha Ojha)

• mgr/dashboard: fix for using ‘::’ on hosts without ipv6 (issue#38575, pr#26751, Noah Watkins)

• mgr: deadlock: _check_auth_rotating possible clock skew, rotating keys expired way too early (issue#23460, pr#26427, Yan Jun)

• mgr: drop GIL in get_config (issue#35985, pr#26613, John Spray)

• mgr/restful: fix py got exception when get osd info (issue#38182, pr#26199, Boris Ranto, zouaiguo)

• mon: A PG with PG_STATE_REPAIR doesn’t mean damaged data, PG_STATE_IN… (issue#38070, pr#26305, David Zafman)

• mon/MgrStatMonitor: ensure only one copy of initial service map (issue#38839, pr#27207, Sage Weil)

• mon: monstore tool rebuild does not generate creating_pgs (issue#36306, pr#25825, Sage Weil)

• mon: scrub warning check incorrectly uses mon scrub interval (issue#37264, pr#26557, Zhi Zhang, Sage Weil, David Zafman)

• msg/async: backport recent messenger fixes (issue#36497, issue#37778, pr#25956, xie xingguo)

• msg/msg_types: fix the dencoder of entity_addr_t (issue#24676, pr#26042, Kefu Chai)

• msgr: should set EPOLLET flag on del_event() (issue#38828, pr#27226, Roman Penyaev)

• msg: should set EPOLLET flag on del_event() (issue#38857, pr#27226, Roman Penyaev)

• multisite: es sync null versioned object failed because of olh info (issue#23842, issue#23841, pr#26358, Tianshan Qu, Shang Ding)

• Object can still be deleted even if s3:DeleteObject policy is set (issue#37403, pr#26310, Enming.Zhang)

• objecter: avoid race when reset down osd’s session (issue#24601, pr#25853, Zengran Zhang)

• os/bluestore: backport new bitmap allocator (issue#24598, pr#26979, Radoslaw Zarzynski, Jianpeng Ma, Igor Fedotov, Sage Weil)

• os/bluestore: bitmap allocator might fail to return contiguous chunk despite having enough space (issue#38761, pr#27312, Igor Fedotov)

• os/bluestore: do not assert on non-zero err codes from compress() call (issue#37839, pr#26544, Igor Fedotov)

• os/bluestore: fix lack of onode ref during removal (issue#38395, pr#26540, Sage Weil)

• os/bluestore: Fix problem with bluefs’s freespace not being balanced when kv_sync_thread is sleeping (issue#38574, pr#26866, Adam Kupczyk)

• os/bluestore: fixup access a destroy cond cause deadlock or undefine (issue#37733, pr#26261, linbing)

• os/bluestore: KernelDevice::read() does the EIO mapping now (issue#36455, pr#25855, Radoslaw Zarzynski)

• osd: backport recent upmap fixes (issue#37968, issue#37940, issue#37881, pr#26127, huangjun, xie xingguo)

• osd: backport recent upmap fixes (issue#38826, issue#38897, pr#27224, huangjun, xie xingguo)

• osd/bluestore: deep fsck fails on inspecting very large onodes (issue#38065, pr#26387, Igor Fedotov)

• OSD crashes in get_str_map while creating with ceph-volume (issue#38329, pr#26900, Sage Weil)

• osd: keep using cache even if op will invalid cache (issue#37593, pr#26078, Zengran Zhang)

• osd/PG.cc: account for missing set irrespective of last_complete (issue#37919, pr#26236, Neha Ojha)

• osd/PrimaryLogPG: fix the extent length error of the sync read (issue#37680, pr#25711, Xiaofei Cui)

• osd/PrimaryLogPG: handle object !exists in handle_watch_timeout (issue#38432, pr#26706, Sage Weil)

• rbd-mirror: update mirror status when stopping (issue#36659, pr#25720, Jason Dillaman)

• rgw: bucket full sync handles delete markers (issue#38007, pr#26192, Casey Bodley)

• rgw: bucket limit check misbehaves for > max-entries buckets (usually 1000) (issue#35973, pr#26946, Matt Benjamin)

• rgw: bug in versioning concurrent, list and get have consistency issue (issue#38060, pr#26548, Wang Hao)

• rgw: check for non-existent bucket in RGWGetACLs (issue#38116, pr#26530, Matt Benjamin)

• rgw: data sync drains lease stack on lease failure (issue#38479, pr#26761, Casey Bodley)

• rgw: fails to start on Fedora 28 from default configuration (issue#24228, pr#26131, Matt Benjamin)

• rgw: feature – log successful bucket resharding events (issue#37647, pr#25738, J. Eric Ivancich)

• rgw: fetch_remote_obj filters out olh attrs (issue#37792, pr#26191, Casey Bodley)

• rgw: fix cls_bucket_head result order consistency (issue#38410, pr#26546, Tianshan Qu)

• rgw: fix radosgw linkage with WITH_RADOSGW_BEAST_FRONTEND=OFF (issue#23680, pr#26332, Nathan Cutler)

• rgw: fix rgw_data_sync_info::json_decode() (issue#38373, pr#26549, Casey Bodley)

• rgw: handle S3 version 2 pre-signed urls with meta-data (issue#23470, pr#25901, Matt Benjamin)

• rgw: ldap: fix LDAPAuthEngine::init() when uri !empty() (issue#38699, pr#27173, Matt Benjamin)

• rgw: multiple es related fixes and improvements (issue#22877, issue#23655, issue#38030, issue#38028, issue#36092, pr#26516, Yehuda Sadeh, Abhishek Lekshmanan)

• rgw multisite: data sync checks empty next_marker for datalog (issue#39033, pr#27299, Casey Bodley)

• rgw: nfs: skip empty (non-POSIX) path segments (issue#38744, pr#27180, Matt Benjamin)

• rgw: only update last_trim marker on ENODATA (issue#38075, pr#26619, Casey Bodley)

• rgw: “radosgw-admin bucket rm … –purge-objects” can hang (issue#38007, issue#38134, pr#26263, J. Eric Ivancich)

• rgw: rgw_file: only first subuser can be exported to nfs (issue#37855, pr#26677, MinSheng Lin)

• rgw: rgwgc: process coredump in some special case (issue#23199, pr#25611, zhaokun)

• rgw: sse-c-fixes (issue#38700, pr#27295, Adam Kupczyk, Casey Bodley, Abhishek Lekshmanan)

• rgw: sync module: avoid printing attrs of objects in log (issue#37646, pr#27030, Abhishek Lekshmanan)

• tools: ceph-objectstore-tool: Dump hashinfo (issue#37597, pr#25722, David Zafman)