Lustre 2.14.0 Changelog

Server support for kernels:

• RHEL 8.3 (4.18.0-240.1.1.el8_3)

Client support for unpatched kernels:

• RHEL 8.3 (4.18.0-240.1.1.el8_3)
• SLES15 SP2 (5.3.18-24.24.1)
• Ubuntu 20.04 (5.4.0-40)

Recommended e2fsprogs version:

• v1.45.6.wc5

Interoperability Support:

• Clients & Servers: Latest 2.12.X, 2.13

Issues fixed between 2.13.0 and 2.14.0

New Feature

• LU-1365: Implement ldiskfs LARGEDIR support for e2fsprogs

Release Notes - Lustre - Version Lustre 2.14.0

Bugs

• [1] - Fix 'Timed out tx' error message
• [2] - Interop 2.4.1<->2.5 failure on test suite sanity-lfsck test_2c
• [3] - sanity test_43: multiop still running
• [4] - sanity-lfsck test_11b: (9) expect lastid1 0x100000000:569
• [5] - sanity-lfsck does not work under ZFS-based DNE mode
• [6] - replay-dual test_22a: Remote creation failed 1
• [7] - replay-single/test_110f test failed Lustre: DEBUG MARKER: replay-single test_110f: FAIL: 1 != 2 after recovery
• [8] - signal vs CP callback race
• [9] - Changes required to allow USE_LU_REF/lu_ref feature enabling and usage
• [10] - libcfs_debug_dumplog calls potentially sleeping function in wrong state
• [11] - non-root user is able to change stripe size for mount point
• [12] - sanity 76 fails on clients with CONFIG_SLAB unset (inode slab grew from 400 to 525)
• [13] - build: Re-add the lustre-devel package.
• [14] - Source rpm does not cleanly build with --rebuild
• [15] - LNET_MAX_IOV hard coded to 256
• [16] - GPF llog_osd_declare_write_rec+0xb6/0x3d0
• [17] - ASSERTION( osd->od_oi_table != NULL && osd->od_oi_count >= 1 ) failed
• [18] - sanity test_133g: timeout during MDT mount
• [19] - sanity-dom test_251: test_sanity failed with 1
• [20] - lnet/utils/lnetconfig/liblnetconfig.c uses sizeof on pointers
• [21] - sanity test 300c fails with 'create 5k files failed'
• [22] - brw rpc reordering causes data corruption when the writethrough cache is disabled
• [23] - Track creates in MDC and llite stats
• [24] - sanity test_133g: ost1 find /proc/fs/lustre/ /proc/sys/lnet/ /proc/sys/lustre/ failed
• [25] - MGC lock leak
• [26] - ptlrpc_set_add_req()) ASSERTION( req->rq_import->imp_state != LUSTRE_IMP_IDLE ) failed
• [27] - racer: mdc_dev.c:1346:mdc_req_attr_set()) uncovered page
• [28] - client hit BUG: unable to handle kernel NULL pointer dereference at 0000000000000028
• [29] - run_rr_alloc() defect: FSNAME should be used instead of lustre
• [30] - preserve PFL/FLR/DoM layout with lfs_migrate
• [31] - lock_count is exceeding lru_size
• [32] - enable large_dir support for MDTs
• [33] - increase default readahead tunables to match common settings
• [34] - "lfs getstripe" on directory does not show default root layout
• [35] - replay-single test 0d fails with 'post-failover df failed'
• [36] - sanity test 255c fails with 'Ladvise test 12, bad lock count, returned 1, actual 0'
• [37] - conf-sanity test_93 osd_handler.c:7132:osd_device_init0()) ASSERTION( info ) failed:
• [38] - ARM servers crashing on MDS startup
• [39] - filefrag/FIEMAP doesn't work for PFL or FLR files
• [40] - doesn't handle logical network interface properly.
• [41] - ASSERTION( tgd->tgd_tot_granted >= ted->ted_grant ) on OSS
• [42] - nodemap_config_set_active()) ASSERTION( config->nmc_default_nodemap ) failed
• [43] - After partial lustre_rmmod, lnet panics on debugfs read
• [44] - Access to invalid semaphore in osd_trunc_unlock_all (ldiskfs)
• [45] - sanity-quota test_2: user create failure, but expect success
• [46] - lctl list_param -R does not display everything
• [47] - possible scheduling with spinlocks held in the quota paths
• [48] - lnetctl peer show hangs for ~2600 clients, ioctl getting E2BIG
• [49] - host_nids_address() fix for MR setup
• [50] - sanity test_76: inode slab grew from 69804 to 69816
• [https://jira.whamcloud.com/browse/LU-12232 - replay-ost-single test 6 fails with space grew after dd: before:13442048 after_dd:13442048
• [51] - Deadlock on LNet shutdown
• [52] - Unable to detect device faults from IB event queue
• [53] - Decrement Health Value even if recovery processing fault
• [https://jira.whamcloud.com/browse/LU-12295 - MDS Panic on DNE2 directory removing
• [54] - Unlink speed needs to be improved in DoM configuration
• [55] - Please fix grammar
• [56] - sanity-quota test 1 fails with 'project quota isn't released after deletion'
• [57] - customise mdsrate tests
• [58] - osp: race around opd_new_connection
• [59] - lli_trunc_sem can lead to a readlock
• [60] - sanity test_230b crash: LBUG of lu_buf_free in mdd_iterate_xattrs
• [61] - sanityn test_47b: create isn't blocked
• [62] - strange breakage in sanityn test 93
• [63] - LustreError: 19435:0:(vvp_io.c:1056:vvp_io_write_start()) LBUG
• [64] - Client unable to mount filesystem with very large number of MDTs
• [65] - udev add/change rule loads zfs module on clients
• [66] - Lustre project PID 32-bit overflow
• [67] - ptlrpcd daemon sleeps while holding imp_lock spinlock
• [68] - usercopy exposure attempt detected in LL_IOC_LOV_GETSTRIPE ioctl
• [69] - truncate to start of uninstantiated PFL component silently fails
• [70] - update_log corruption
• [71] - osd_ios_lf_fill() returns 0 on some error paths.
• [72] - sanity-hsm test_90: FAIL: requests did not complete
• [73] - Support RHEL 8.1
• [74] - bad sequence output on OST
• [75] - correct fast read & strided readahead interaction
• [76] - Restore performance enhancement from LU-11413
• [77] - High kworker CPU usage (osc_grant_work_handler) on IDLE connections
• [78] - osd_attr_set() fails with ENOSPC on linux-4.18
• [79] - osd_oxc_get(): Shrink rcu critical section to not include EXIT
• [80] - FIxing conf-sanity.sh test_21d (zfs)
• [81] - large-lun test_2 should check / warn if OSTSIZE is too small (zfs)
• [82] - Fast ENOSPC on direct I/O
• [83] - sanity-quota test_4a: create after timer expired, but expect EDQUOT
• [84] - racer test_1: Invalid layout: The component end must be aligned by the stripe size
• [85] - Readahead can trigger an assertion failire in cl_io_read_ahead
• [86] - sanity-sec "error: set_param: setting /sys/fs/lustre/mdt/lustre-MDT0000/identity_upcall=NONE: wrote only 1"
• [87] - sanity test_807: f807.sanity expected blocks: 4103, got: 2051
• [88] - crash in osd_object_delete at end of sanity
• [89] - sanity: test 811 fail with "MDD orphan cleanup thread not quit"
• [90] - parallel readahead needs to be optimized at high number of process
• [91] - sanity-lfsck test 36a fails with '(N) Fail to resync /mnt/lustre/d36a.sanity-lfsck/f2'
• [LU-12759 - parameter grant_shrink gets reset to 1 after client reconnects
• [92] - sanity test 103a syntax error using version_code
• [93] - readcache_max_filesize in 2.12?
• [94] - Sanity quota test fixes for zfs
• [95] - conf-sanity test 32c fails with ‘mv remote dir failed’
• [96] - sanity-sec test 12 fails with 'nodemap_add failed with 1'
• [97] - sanity test_272a crashes with SSK
• [98] - sanity-quota: test 3 failure "write success, but expect EDQUOT"
• [99] - kernel update [RHEL8.0 4.18.0-80.11.2.el8_0]
• [100] - NULL pointer dereference IP: ptlrpc_disconnect_and_idle_import
• [101] - recovery-small/110k gets stuck at client umount
• [102] - lustre_swab_fiemap needs to check fm_mapped_extents
• [103] - LBUG: (niobuf.c:350:ptlrpc_register_bulk()) ASSERTION( !(desc->bd_registered && req->rq_send_state != LUSTRE_IMP_REPLAY) || mbits != desc->bd_last_mbits )
• [104] - gcc9 support
• [105] - basename in test-framework to support capital alphabets
• [106] - Project quotas: users can change project IDs
• [107] - FLOCK request can be processed twice during resend
• [108] - lustre_lwp_add_conn can't find lwp device
• [109] - could not delete orphan [0x2001ac5ff:0x3a30:0x0]: rc = -2
• [110] - growing a PFL file with last stripe as -1 fails
• [111] - general protection fault: 0000 RIP: keys_fill
• [112] - Allow "true/false" or "on/off" for audit_mode on nodemap
• [113] - LustreError: 82937:0:(ldlm_lib.c:3268:target_bulk_io()) @@@ truncated bulk READ 0(270336)
• [114] - sanity test 160f fails with ‘mds1: User cl6 not registered’
• [115] - sanity-hsm test 9A fails with “uuid Dumping not found in agent list on mds1”
• [116] - Do not assume peers are MR capable
• [117] - SSK regression in 2.12.3
• [118] - review-dne-selinux sanity test_185a: lu_object_attr()) ASSERTION( ((o)->lo_header->loh_attr & LOHA_EXISTS) != 0 ) failed
• [119] - building lustre utils on ppc64le fails with error format expects llu but argument is __u64
• [120] - Failing to create a properly sized IB queue pair
• [121] - Setting a LOV EA can access or change outside allocated buffer
• [122] - (mdt_open.c:312:mdt_prep_ma_buf_from_rep()) ASSERTION( ma->ma_lmv == ((void *)0) && ma->ma_lmm == ((void *)0) ) failed
• [123] - Fix source specified remote destination send case
• [124] - build: lbuild script should not continue if it's missing ed
• [125] - recovery-small test_136: crash in sec2target_str() with review-dne-selinux-ssk
• [126] - make newwiretest adds strings, not connected to changes
• [127] - Migration corrupts xattrs
• [128] - Avoid sending PageSlab pages through tcp stack
• [129] - Multipath path flapping issue
• [130] - extend_recovery_timer assertion
• [131] - LL_IOC_GETOBDCOUNT return wrong MDT counts
• [132] - AST replies are dropped when servers are non-MR, clients and routers are MR
• [133] - oops in target_handle_connect()
• [134] - create striped directory on specific MDTs not working
• [https://jira.whamcloud.com/browse/LU-12965 - class_import_put()) ASSERTION( _v >= 0 && _v < 0x5a5a5a5a ) failed: value: 1515870810
• [135] - sanity test 80 silently fails to get sync_on_lock_cancel parameter
• [136] - KASAN: slab-out-of-bounds in mgs_wlp_lcfg+0xb3/0x4a0 [mgs]
• [137] - make sure we reset quota limits for all test ID/users
• [138] - Running lfsck with dry-run (-n) should not move orphan obhect to lost+found
• [139] - Make a maximal transfer size check accurately.
• [140] - sanity test_240: failed to mount client
• [141] - gcc8 compile warning on string overflow in name_create_mdt
• [142] - check_routers_before_use causes LNet to hang indefinitely if any router is down
• [143] - sanity-lnet stops all test suites from running
• [144] - sanity test 56ra fails with ' '/usr/bin/lfs find -size 0 -type f /mnt/lustre/d56ra.sanity' should send 12 glimpse RPCs to OST'
• [145] - sanity test_60g: usleep: command not found
• [146] - LNet Health: Peer net health not considered
• [147] - LNet Routing: gateway NI not selected properly
• [148] - LNet Routing: route consolidation
• [149] - LNet Discovery: toggling discovery on/off is not handled properly
• [150] - LNet Routing: asym routing is not working for multi-hop routing
• [151] - LNet MR: fix remote peer NI selection
• [152] - BUG: KASAN: stack-out-of-bounds in strchr+0x2d/0x50
• [153] - BUG: KASAN: slab-out-of-bounds in string_nocheck+0xd1/0x180
• [154] - LNet: bug in peer deletion
• [155] - sanity-selinux test_21b Error: 'touch (5)'
• [156] - Project quotas: force to ignore quota flags =8 on quota exceeded
• [157] - LNet: Handle shutdown properly
• [158] - conf-sanity test 87 call to umount_ldiskfs() fails
• [159] - MDS kernel BUG at ldiskfs/htree_lock.c:429!
• [160] - sanity-pcc test_1c: Failed to start copytool ost8
• [161] - osd_fid_lookup()) ASSERTION( fid_is_sane(fid) || fid_is_idif(fid) ) failed: [0x0:0x68:0x0]
• [162] - Check permissions when accessing changelogs
• [163] - runtests test 1 hangs when unmounting an OST
• [164] - osp_sync_thread()) ASSERTION( thread->t_flags != SVC_RUNNING ) failed: 308 changes, 111 in progress, 0 in flight
• [165] - mdd_orphan_destroy loop caused by compatibility issue on upgrades to 2.11 or later
• [https://jira.whamcloud.com/browse/LU-13071 - LNet Health: reduce log severity
• [166] - Cleanup xattr checking
• [167] - Enable lgss_keyring debug traces
• [168] - The server highest transno can be lost, client reports 'went back in time'
• [169] - sleeping function in target_recovery_overseer
• [170] - fix lfs_migrate -p for file with an existing pool
• [171] - include lbuild-{fc,rhel,sles} to SIGNATURE
• [172] - osd_trans_exec_check()) lustre-OST0001: opcode 0: used 8, used now 8, reserved 1 LBUG
• [173] - incorrect used_mb in max_cached_mb
• [174] - sanityn test_47b: create must fail
• [175] - ll_set_inode()) Can not initialize inode warning
• [176] - Eviction during ll_open_cleanup()
• [177] - llog_cat_id2handle()) error opening log id rc = -2 osp_sync_thread()) ASSERTION( thread->t_flags != SVC_RUNNING )
• [178] - always check object existence after mo_xattr_get()
• [179] - sanity-sec test_30b: touch skn
• [180] - Wrong job id parsing in cfs_get_environ()
• [181] - lustre-initialization crashed in common_file_perm() on SLES12
• [182] - ll_update_lsm_md() may deadlock
• [183] - a race between glimpse and lock cancel is not handled correctly
• [184] - Partial writes on multi-client strided files
• [185] - sanity-selinux test_21a and test_21b: sepol update in case of combined MGS/MDS
• [186] - (layout.c:2121:__req_capsule_get()) @@@ Wrong buffer for field 'niobuf_inline' (7 of 7) in format 'LDLM_INTENT_OPEN', 0 vs. 0 (server)
• [187] - User process segfaults since 2.13 client upgrade
• [188] - Cleanup layout checking
• [189] - LNet Health: increase transaction timeout
• [190] - sanity-lnet does not cleanup after itself on test failure
• [191] - llapi layout llc_pattern is not set properly on DoM component
• [192] - Function llapi_get_version_string fails
• [https://jira.whamcloud.com/browse/LU-13154 - Interop: sanity-quota test 66 fails with 'nonroot user should fail to set projid'
• [193] - sanity-selinux test_21a test_21b: c0.sepol not updated on mds1 after 20 secs
• [194] - migrate symlink with target name length > 59 cause crash
• [195] - osd-zfs can leak index backup entries
• [196] - sanity-hsm test 70 timeout
• [197] - sanity test_65i hung: RIP: 0010:mdc_read_page+0x14f/0x9b0 [mdc]
• [198] - MSG_RESENT can be improperly cleared.
• [199] - sanity test_812b: FAIL: lfs quota should succeed
• [200] - Client panic "Freechain corrupt"/"Redzone Overwritten"
• [201] - sanity-lfsck/23b fails often with ZFS
• [202] - sanity-quota convert minutes to seconds to sleep
• [203] - kiblnd_fmr_pool_map error on the AARCH64 with 64k pages
• [204] - MAP_POPULATE hangs with Linux 5.4
• [205] - Linux 5.4 ldiskfs build (drop: ext4-remove-truncate-warning.patch)
• [206] - conf-sanity test_112: problem creating f112.conf-sanity.0 on OST0000
• [207] - sanity test_129: current dir size 4096, previous limit 20480
• [208] - I/O can fail with -EBADF if fresh client has no connection to OST
• [209] - osp_sync_interpret() LBUG upon OST disk failure
• [210] - Interop: sanityn test 104 fails with '/mnt/lustre/f104.sanityn mtime (1580286855:1580286853) diff'
• [211] - Remove inode mutex for DIO reads
• [212] - SUSE 15 SP2 client build fix
• [213] - Build failure with gcc8 -Werror=implicit-fallthrough=
• [214] - ldiskfs for 5.4.x post 5.4.6
• [215] - Kernel NULL pointer dereference in lustre_msg_set_conn_cnt()
• [216] - parallel-scale-nfsv3: FAIL: setup nfs failed on RHEL 8.1
• [217] - add bash completion for lfs
• [218] - sanityn 16a FAIL: fsx with O_DIRECT failed.
• [219] - write access to an mmapped file over soft quota limit causes sigbus
• [220] - Flock unlock request resend caused a memory leak.
• [221] - sanity test 160j fails with 'read changelog failed'
• [222] - ldiskfs/namei.c:3310 ldiskfs_orphan_add+0x11e/0x2a0 [ldiskfs]
• [223] - Fix wrong size in lnet_net_append_cpts
• [224] - osd_otable_it_next() can miss spin_unlock()
• [225] - Handle possible failure from name_create_mdt
• [226] - sanity/177 leaves threads blocked in libcfs_debug_dumplog()
• [227] - crash at lu_object_find() in mdt_lvbo_fill()
• [228] - Fix size checking issue in lov
• [229] - Lustre : lfs1-MDT0000: small buffer size
• [230] - "(lu_ref.c:257:lu_ref_del()) ASSERTION( 0 ) failed" triggered by jbd2/ldlm tasks, with recent master configured with USE_LU_REF defined
• [231] - "(lu_ref.c:257:lu_ref_del()) ASSERTION( 0 ) failed" triggered by lu_ref_del() in osc_ldlm_glimpse_ast(), because no corresponding lu_ref_link posted, with recent master configured with USE_LU_REF defined
• [232] - Multiple "(cl_object.c:962:cl_env_percpu_put()) ASSERTION( cle == &cl_env_percpu[cpu] ) failed" cases triggered by cl_env_percpu_put() in osc_lock_upcall() context, with recent master configured with USE_LU_REF defined
• [233] - ln_push_target can be overwritten leading to incorrect lnet_peer objects.
• [234] - improve racer cleanup
• [235] - dom-performance 4 defects
• [236] - skip conf-sanity.sh 87, 115 tests if FILSET set
• [237] - run_ior() parameters order fix
• [238] - Potential infinite loop in push target resize code
• [239] - Potential deadlock in lnet_peer_discovery
• [240] - Asymmetry in discovery push vs reply handling
• [241] - sanity.sh:test_20() does not take DoM component into account
• [242] - ha.sh 1a7c352e97 fix: wait $LFSCK_BG_PID properly
• [243] - a performance regression on the single stream write
• [244] - mballoc should not skip uninitialized-on-disk groups
• [245] - Readahead doesn't work well for non-stride SSF
• [246] - wrong cpt malloc rotor handling leads to oops
• [247] - statfs isn't work properly with MDT statfs proxy
• [248] - lfs setstripe to deactivated OST does not return
• [249] - sanity test_56ob: lfs find /mnt/lustre/d56ob.sanity -mtime 1y' wrong: found 0, expected 1
• [250] - NULL pointer dereference at vvp_page_delete+0x14/0x140
• [251] - kernel update [SLES12 SP4 4.12.14-95.48.1]
• [252] - Statahead should skip the entry while fid change
• [253] - Test duration not reset when using ONLY_REPEAT test parameter
• [254] - update wirecheck for missed values
• [255] - adler32 wrapper in libcfs
• [256] - lctl conf_param hung on the MGS node
• [257] - striped directory doesn't support lookup
• [258] - Race between message finalize and discovery reply processing
• [259] - spin_lock in after_reply() eat up most of cpu
• [260] - lnet may be trying to use deleted routes leading to errors kiblnd_rejected(): 10.0.10.212@o2ib7 rejected: consumer defined fatal error
• [261] - Wrong verify flags for writing in ll_file_aio_write
• [262] - lfs mirror extend does not take file layout into account
• [263] - potential dead loops on short writing
• [264] - interop b2_12/master: lustre-initialization failed: Unrecognized mount option "localrecov"
• [265] - allow over reservation to cover current read
• [266] - aborted file create may leave unattached inodes on MDS.
• [267] - too many namespace LFSCK repair the shard messages
• [268] - sanity test_252: Invalid number of mdtlov clients returned by /usr/sbin/lr_reader
• [269] - ‘lfs mirror read’ does not write to output file
• [270] - tgt_cancel_slc_locks()) ASSERTION( lock->l_client_cookie != 0 )
• [271] - processing of update logs couldn't delete empty plain llogs
• [272] - Data corruption during IOR testing with DoM files and hard failover
• [273] - Some minor typos crept into gnilnd
• [274] - Fix cleanup_netns in sanity-lnet.sh
• [275] - unable to migrate mirrored files
• [276] - "run 'lfs migrate -m 1 -c 1 -H 3 dir1' to finish migration" is broken
• [277] - "lfs migrate" on DoM component clobbers LOV EA FID
• [278] - "lfs df -v" prints too many MDTs
• [279] - client login node crash: kernel:LustreError: 190:0:(lov_io.c:121:lov_io_sub_init()) ASSERTION(is_index_within_mirror(lov, index, lio->lis_mirror_index) ) failed: [0x20004f178:0x2f6c:0x0]iot = 6, index = 0, mirror = 0
• [280] - Rework __vfs_getxattr/__vfs_setxattr/__vfs_removexattr support
• [281] - Rework preferred IO size
• [282] - test-frameworkto preserve MODOPTS_LIBCFS
• [283] - lfs mirror delete "usage" and manpage do not match
• [284] - rename may miss revoking LOOKUP lock to cause stale dentry on client
• [285] - ASSERTION( i < 1000 ) failed
• [286] - recovery-small test_140b: FAIL: recovery took too long 4362 20
• [287] - NULL dereference in lnet_health_check lnet_incr_hstats
• [288] - HSM client disconnected and fail to reconnect to server
• [289] - use correct obd index in print_failed_tgt()
• [290] - LNet routing: wrong gw ni may be selected to reach undiscovered peer
• [291] - MDT0000 remount in recovery 40 hours
• [292] - Interop 2.12 <-> master sanity test_27A: lfs setstripe: setstripe error for '/mnt/lustre': Unknown error 524
• [293] - BUG: unable to handle kernel NULL pointer dereference in class_exp2cliimp
• [294] - truncate deadlock with DoM files
• [295] - sysfs ping write creates a flood-ping situation that could not be normally stopped
• [296] - LNet: discovery can glitch
• [297] - lnet: routing aliveness might be stale
• [298] - page discard of non-composite file does not set mirror info
• [299] - Lustre failover fails when SRPC enabled
• [300] - vvp_set_pagevec_dirty / vvp_page_completion_write lock ordering appears to trigger RCU stalls
• [301] - LNet: Force full discovery
• [302] - LNet: peer update adjustment on discovery toggle
• [303] - sanity test_33h: MDT index mismatch 5 times
• [304] - tests sanity/32* fail
• [305] - readahead thread breaks read stats in jobstats
• [306] - vvp_set_pagevec_dirty()) ASSERTION( page->mapping ) failed: mapping must be set
• [307] - sanity test 56w fails with '/usr/bin/lfs_migrate -y -c 7 /mnt/lustre/d56w.sanity/file1 failed '
• [308] - client UUID is truncated
• [309] - crash in sanity test 160j
• [310] - MDS 2.12.4 ASSERTION( top->loh_hash.next == ((void *)0) && top->loh_hash.pprev == ((void *)0) ) failed
• [311] - expand inode if possible for project quota
• [312] - a performance regression on directory creation
• [313] - mdt.mdt_mds_mds_conns is not counted upon MDT stop and start
• [314] - struct sepol_downcall_data is badly formed
• [315] - sanity test_69: read succeeded, expect -ENOENT
• [316] - Files truncated/corruption due to lfsck
• [317] - Some LASSERTF are missing terminating newlines
• [318] - Application hang and kernel "divide error" in ll_readpage
• [319] - hsm.actions file is broken on RHEL 8.2
• [320] - 'make rpms' fails on centos8.1 client
• [321] - o2iblnd: convert LASSERT (tx->tx_sending > 0); to proper error handling
• [322] - Do not treat repo ml kernels as 'special' just map them to RHEL
• [323] - LU-11023 (quota pools) breaks building with gcc9
• [324] - lfs mirror delete wrong error message
• [325] - socklnd: wrong NID to interface mapping
• [326] - sanity test_39r: atime on client != ost
• [327] - sanity-selinux test_5: sec context seen from 1st mount point is not correct
• [328] - LU-9859 libcfs: add support for Xarray breaks build on 3.10.0-1062.9.1.el7.x86_64.debug
• [329] - Fix growing message buffer
• [330] - LustreError: 30166:0:(service.c:189:ptlrpc_save_lock()) ASSERTION( rs->rs_nlocks < 8 ) failed
• [331] - Intermediate routers modify the final destination NID
• [332] - Poor performance with routed clients to multi-rail servers when single server interface fails
• [333] - MDT stuck in WAITING, abort_recov stuck too
• [334] - lctl --device MGS llog_catlist doesn't list all config files.
• [335] - replay-single test_117: LBUG: ASSERTION( atomic_read(&obd->obd_req_replay_clients) == 0 ) failed
• [336] - Client dead lock leads to eviction from MDS (selinux is enabled)
• [337] - sanityn test_106a test_106c: supported stx_attributes: got ?, expected 30
• [338] - Add -D option back to lfs_migrate
• [339] - Various data corruptions possible in lustre.
• [340] - Route status can be set incorrectly via lnet_notify()
• [341] - osd_destroy()) ASSERTION( !lu_object_is_dying(dt->do_lu.lo_header) ) failed
• [342] - Ignore quota on slave MDTs during creating striped directory
• [343] - LNet: crash on bad user space parameters
• [344] - sanityn test_80b: ASSERTION( lo->ldo_dir_stripe_count > 0 ) failed
• [345] - ptlrpc_pinger_main is stuck in endless loop
• [346] - quiet compiler warnings on arm7l
• [347] - LNetError: 14769:0:(o2iblnd.h:1003:kiblnd_queue2str()) LBUG
• [348] - quota pools qunit sorting doesn't work
• [349] - large allocations in osd_bufs_get() failing
• [350] - lfs migrate on non-existing file returns exit code 0
• [https://jira.whamcloud.com/browse/LU-13686 - lctl pool_add returns error randomly
• [351] - interop: "lfs find" prints spurious "No such file or directory" error messages
• [352] - Handle in-tree executed HSM copytools correctly
• [353] - lfs mirror ... /mnt/lustre/.lustre/fid/$FID causes CWARN() about invalid FID formats
• [354] - MDS slow/hung threads at mdt_object_local_lock
• [355] - lfs getstripe should avoid opening regular files
• [356] - short io for AIO
• [357] - sanity test_230o: dir split not finished
• [358] - sanity-lnet test_205: @@@@@@ FAIL: Should have failed
• [359] - 'lfs mkdir -i -1' may not initialize target MDTs correctly
• [360] - Typo in libssl dependency
• [361] - Flaw in MR Routing Algorithm
• [362] - LNet Router: bug in the predetermined gateway case
• [363] - lov tgt 36 not cleaned! deathrow=0, lovrc=1
• [364] - "lfs find --pool" does not work for PFL files
• [365] - lnetctl: inconsistent global show
• [366] - list_add corruption messages seen in a stack trace seen while trying to mount OST
• [367] - RHEL 8.2: VM_FAULT_RETRY is defined in mm_types.h instead of mm.h
• [368] - "lfs fid2path" can't match the rootpath correctly
• [369] - report client stats sumsq
• [370] - Do not set preferred NI for MR peer
• [371] - Provide an selinux_is_enabled for newer kernels
• [372] - lctl llog_print segfault if no log name supplied
• [373] - LNet lnetctl: peer add doesn't work as specified in the manual
• [374] - sanity test_160j: FAIL: read changelog failed
• [375] - GSS-based authentication fails on CentOS/RHEL 7.8
• [376] - sanity-dom sanityn_test_20 fails with '1 page left in cache after lock cancel'
• [377] - ptlrpc_invalidate_import()) lsrza-OST0000_UUID: Unregistering RPCs found (0). Network is sluggish? Waiting them to error out.
• [378] - lp_dc_error is not cleared when new discovery ping is sent as response to incoming push
• [379] - ldiskfs_mb_mark_diskspace_used:3472: aborting transaction: error 28 in __ldiskfs_handle_dirty_metadata
• [380] - Wrong memory copy in tgt_pages2shortio
• [381] - pjdfstest script should not write outside of /tmp
• [382] - llmount.sh cannot start Lustre with project quota enabled
• [383] - Build failure libcfs needed by snmp
• [384] - socklnd: NID to interface mapping issues
• [385] - Capabilities are not effective
• [386] - PFL file lost during recovery
• [387] - can not install lustre-client-utils on ubuntu 20.04
• [388] - ptlrpc_expired_set() needs to reschedule from time to time
• [389] - sanity test 230q fails with 'stripe hash crush,migrating != crush'
• [390] - fix compatibility for LL_IOC_MDC_GETINFO
• [391] - reuse same cl_dio_aio for one IO
• [392] - lctl show_route prints incorrect route aliveness
• [393] - Remove use of "-o" for lfs setquota
• [394] - Kernel crash on: lfs quota -u $(( (1<<32) -1))
• [395] - move iov iter forward by ourself
• [396] - build: missing dependency check on binutils-devel
• [397] - ladvise: buffer allocated for ioctl is never released
• [398] - Reference leak in check for mr_forwarding
• [399] - IOR data corruption detected during automated lnet fofb testing
• [400] - don't call aio_complete() in lustre upon errors
• [401] - release invalid dentries proactively on client
• [402] - return 0 if no error is found in mdt_path_current()
• [403] - Avoid atomic operation while bitmap prefetch
• [404] - lfs mirror extend command ignores flags
• [405] - HSM: hsm_actions are not processed after MDS failover
• [406] - Client blocked in lstat()
• [407] - lhsm_posix with lt- prefix when build from source tree with --disable-shared
• [408] - ofd_attr_set and ofd_destroy do not properly handle object removal
• [409] - it_lock_bits should be bit-wise tested
• [410] - lnet-selftest fails due to clients and server lists are empty
• [411] - copy_file_range() compat definition lacks proper headers
• [412] - t-f defect: MODOPTS_<module> ignored on FAILOVER=HARD mode
• [413] - autogen.sh is missing from the tarball generated by "make dist"
• [414] - conf-sanity test 53a and 53b fail with ‘/usr/lib64/lustre/tests/functions.sh: line 152: var: invalid indirect expansion’
• [415] - Tests fail/hang with “can't notify - lge_glbl_data is not set”
• [416] - kiblnd can continue attempting to reconnect indefinitely.
• [417] - 'ls -R' fails on encrypted dir without key
• [418] - tests should be consistent in use of "truncate"
• [419] - test_184c has possible infinite loop.
• [420] - rmdir should release inode on Lustre client
• [421] - Failing to send lock callbacks keeps a CPU busy
• [422] - seq_file next function must change *pos
• [423] - livelock is possible in distribute_txn_commit_thread()
• [424] - Don't re-enqueue glimpse lock on read
• [425] - mdt_lock_handle_fini()) ASSERTION( !lustre_handle_is_used(&lh->mlh_reg_lh) ) failed
• [426] - BL AST vs failed lock enqueue race
• [427] - ldlm_flock_deadlock() ASSERTION(req != lock) failed
• [428] - osc_extent_make_ready()) ASSERTION( last_oap_count > 0 ) failed
• [429] - osc_object_ast_clear()) ASSERTION( lvb != ((void *)0) ) failed
• [430] - tests: sanityn test_104 incorrect MDT times printed
• [431] - raise CAP_NET_BIND_SERVICE before calling rdma_resolve_addr()
• [432] - BUG at ldiskfs/htree_lock.c:892
• [https://jira.whamcloud.com/browse/LU-14010 - dkms-deb client package missing dependant packages
• [433] - new layout incompletely initialized in lod_layout_convert()
• [434] - libcfs: use atomic64_t for libcfs_kmemory
• [435] - ofd_inconsistency_verification_main use after free on shutdown.
• [436] - Client recovery statemachine hangs in recovery disconnected during lock reply
• [437] - long time between reconnects
• [438] - lbuild fail to build MOFED 5.1

   * [LU-14037] - lu_device_fini()) ASSERTION( atomic_read(&d->ld_ref) == 0 )
   * [LU-14039] - Set LA_TYPE while working on osp-mdt device
   * [LU-14042] - LBUG LustreError: 7775:0:(lov_io.c:475:lov_io_slice_init()) ASSERTION( obj->lo_lsm != ((void *)0) ) failed:
   * [LU-14045] - Fix O_DIRECT and encrypted files
   * [LU-14049] - ofd_access_log_reader thread leak
   * [LU-14050] - ofd_access_log_reader off by one error in alr_sort_and_print_thread()
   * [LU-14051] - ofd_access_log_reader should flush batch file in the sort and print thread
   * [LU-14053] - ofd_access_log_reader may fail to notice that consumer exited
   * [LU-14057] - Quiet innacurate log message on server side
   * [LU-14061] - LU-12514 (lustre_tgt mount type) breaks scripts that used 'mount -t lustre' and expected targets
   * [LU-14069] - OBD_FAIL_LDLM_CANCEL_BL_CB_RACE is buggy in ldlm_handle_cp_callback
   * [LU-14070] - consider recovering case in tgt_brw_unlock()) assertions
   * [LU-14072] - client evicitions with DIO
   * [LU-14080] - gnilnd updates for 5.3+ kernel changes
   * [LU-14082] - Use l_req_mode while lock cleanup
   * [LU-14083] - Don't overwrite KBUILD_EXTRA_SYMBOLS
   * [LU-14095] - Multiple tests crash with “ASSERTION( rsi->h.cache_list.next == ((void *)0) ) failed “
   * [LU-14097] - conf-sanity/91 fails on local setup
   * [LU-14100] - gnilnd minor build fix
   * [LU-14102] - sanity/160l FAIL: MTIME changelog not recorded
   * [LU-14105] - lfsck shouldn't LBUG() on disk data
   * [LU-14109] - export --backup option is missing from lnetctl man page
   * [LU-14115] - Project Quota broken on Stripe directory
   * [LU-14116] - gss: undefined reference to 'krb5int_enc_des3'
   * [LU-14121] - EACESS Permission denied for slurm log files when using nodemap with admin=0
   * [LU-14125] - client starved for grant but OST has plenty of free space
   * [LU-14128] - truncate does not set file size correctly in corner case
   * [LU-14130] - Skip 0 row in rpc_stats procfs
   * [LU-14131] - reduce credits for overwritting
   * [LU-14136] - lfs mirror extend prints confusing error messages
   * [LU-14143] - SEEK_HOLE returns -ENXIO if file ends at components boundary
   * [LU-14149] - FIEMAP should set FIEMAP_EXTENT_DATA_ENCRYPTED for fscrypt files
   * [LU-14151] - GSS context initialization fails on RHEL8.2 client
   * [LU-14155] - Avoid easize set to 0
   * [LU-14158] - lfs changelog do not display old changelog after changelog_catalog  wrapped arround
   * [LU-14159] - gcc8 warning on kthread_run calls
   * [LU-14161] - conf-sanity/91 may fail due to rhashtable restart
   * [LU-14172] - DIR Stat performance regression in striped dir
   * [LU-14181] - sanity test_64e fails with SHARED_KEY on RHEL/CentOS 8.2
   * [LU-14185] - alloc fails for OSP obj xattr cache
   * [LU-14186] - fallocate does not extend uninitialized PFL components
   * [LU-14187] - LU-14134 causes many racer timeouts
   * [LU-14191] - setstripe: cannot create composite file: Numerical result out of range
   * [LU-14192] - overstriping is not checked for existence in the 1st component
   * [LU-14198] - Several dist-* make-targets broken when building with uid > 2^21
   * [LU-14199] - sanity-selinux test 21a  fails with 'client mount without sending sepol should be refused'
   * [LU-14205] - RHEL8.3: sanity test 398c fails with 'fio write error'
   * [LU-14208] - sanity-hsm test 55 fails with 'request on 0xM:0xN:0x0 is not FAILED on mds1'
   * [LU-14214] - incorrect locking in ofd_object_fallocate()
   * [LU-14227] - sanity-lnet test 21 missing space leading to invalid command
   * [LU-14263] - Kerberos ticket renewal not processed by lustre
   * [LU-14267] - osd_ldiskfs_write_record(): do not update in-bh inode every time
   * [LU-14281] - sanity-sec test 56 fails with 'filefrag /mnt/lustre/d56.sanity-sec/f56.sanity-sec failed'
   * [LU-14283] - sanity-sec test 52 crashes due to ‘divide error: 0000 [#1] SMP PTI’
   * [LU-14286] - fallocate() does not zero on-disk data
   * [LU-14296] - really raise CAP_NET_BIND_SERVICE before calling rdma_resolve_addr()
   * [LU-14298] - sanity-sec test_44 fails on aarch64 client
   * [LU-14299] - sanity-quota test 61 fails with 'write failed, expect succeed'
   * [LU-14301] - EOPNOTSUPP, ENOTSUPP, ENOTSUP confusion
   * [LU-14303] - parallel-scale test rr_alloc fails with 'failed while setting qos_threshold_rr & creat_count'
   * [LU-14306] - sanity-sec test_52: BUG: Bad rss-counter state
   * [LU-14307] - sanity test 60g crashes with ‘unable to handle kernel NULL pointer dereference at 0000000000000020’ in lquota_lqe_debug
   * [LU-14312] - Interop: sanity test 272b fails with 'failed to migrate to the new composite layout'
   * [LU-14317] - building lustre on arm64 hit warning: format '%llu' expects argument of type 'long long unsigned int'
   * [LU-14324] - fsx fallocate detection is buggy
   * [LU-14326] - sanity-dom test_fsx: crash in osc_extent_make_ready()
   * [LU-14337] - return stripe_count=1 instead of 0 for DoM files
   * [LU-14339] - Add a option %H to avoid the jobid too long
   * [LU-14355] - LustreError: 3782:0:(sec.c:411:import_sec_validate_get()) import ffff8dff56545800 (CONNECTING) with no sec
   * [LU-14356] - mkfs.lustre exited with 22 on Ubuntu 20.04.1 aarch64
   * [LU-14389] - crash in lov_delete_composite() with racer+migrate
   * [LU-14424] - write performance regression in Lustre-2.14.0-RC1

• • Test

   * [LU-13586] - check quota pools with PFL and SEL
   * [LU-13810] - Check OST pool quota hard limit at file with wide striping
   * [LU-13817] - Make lfs quota -uv --pool more verbose

• • New Feature

   * [LU-3606] - Implement fallocate() support for ldiskfs Lustre
   * [LU-10997] - Ubuntu 18 support
   * [LU-11023] - OST Pool Quotas
   * [LU-11025] - DNE3: directory restripe
   * [LU-11200] - Centos 8 arm64 server support
   * [LU-12036] - Add option to create new OSTs inactive
   * [LU-12275] - Client-side file data encryption
   * [LU-12461] - Contribute epython scripts to aid crash dump analysis
   * [LU-13180] - Reserve bit for RDMA-only memory RPC
   * [LU-13385] - lnetctl: calculate Lustre traffic service-id 
   * [LU-13946] - OpenZFS 2.0 Compatibility

• • Task

   * [LU-12861] - Add and use scnprintf() where using snprintf() is not handling return correctly	
   * [LU-12904] - Support for linux kernel version 5.4
   * [LU-13006] - need test for /sys/fs/lustre/jobid_this_session
   * [LU-13238] - add OFD layer object access logs
   * [LU-13255] - Abstract code for wait-with-periodic-warning
   * [LU-13359] - change lfs quota --pool to print used based only on OSTs from the pool
   * [LU-13376] - add batching to ofd_access_log_reader
   * [LU-13479] - "could not allocate data integrity bioset" on centos7
   * [LU-13904] - Support non-modular builds
   * [LU-14034] - test-framework and sanity-hsm use libtool nonsensically

• • Improvement

   * [LU-1538] - cleanup test scripts
   * [LU-2225] - sanity/27m to poll for specific state
   * [LU-4198] - Improve IO performance when using DIRECT IO using libaio
   * [LU-8465] - parallel e2fsck performance at scale
   * [LU-9019] - Migrate lustre to standard 64 bit time kernel API
   * [LU-9091] - Replace lprocfs_str_with_units_to_s64 with string_get_size()
   * [LU-9114] - Make MDS (And other server threads?) hog CPU less
   * [LU-9441] - Use kernel threads in predictable fashion to confine OS noise
   * [LU-9555] - "df /path" and "lfs df /path" should return projid-specific values
   * [LU-10235] - mkdir should check for directory existence on client before taking write lock
   * [LU-10428] - LNet events should generated without resource lock held
   * [LU-10447] - Replace use of $SETSTRIPE and $GETSTRIPE with '$LFS setstripe' and '$LFS getstripe' in test suites
   * [LU-10467] - use standard linux wait_event macros
   * [LU-10810] - SEEK_HOLE and SEEK_DATA support for lseek
   * [LU-10934] - integrate statx() API with Lustre
   * [LU-11310] - support for SLES 15
   * [LU-11365] - test-framework::add() should load ldiskfs
   * [LU-11505] - improve functions() run mpi tests with ability to set the composite layout
   * [LU-11607] - Reduce repeated function calls in Lustre test suites
   * [LU-11623] - Allow caching of open-created dentries
   * [LU-11631] - symlink migration should not create remote inodes
   * [LU-11668] - mdd_parent_fid()) ASSERTION( (((mdd_object_type(obj)) & 00170000) == 0040000) ) failed
   * [LU-11963] - Add nonrotational flag to obd_statfs
   * [LU-11971] - Send file creation time to clients 
   * [LU-12043] - improve Lustre single thread read performances
   * [LU-12071] - bypass pagecache for large files
   * [LU-12133] - sanityn test_35() defect
   * [LU-12137] - update client to use iterate_shared
   * [LU-12222] - LNet should select loopback NI when possible
   * [LU-12280] - Provide option to disable soft quota expiration
   * [LU-12303] - Use lnet_health_sensitivity for restoring health for each lnet_recovery_internal
   * [LU-12332] - Add a liblustreapi call for IOC_MDC_GETFILEINFO
   * [LU-12353] - optimizations for ldiskfs quota updates
   * [LU-12355] - Support for linux kernel version 5.0
   * [LU-12400] - Support for linux kernel version 5.2
   * [LU-12410] - Refactor lnetctl peer nid handling to utilize nidstrings library
   * [LU-12477] - Remove obsolete config checks
   * [LU-12518] - improve Lustre unaligned IO read performances
   * [LU-12521] - print_instance() incorrect if fsname contains a dash
   * [LU-12542] - LDLM improvements form linux lustre client work
   * [LU-12546] - add option to abort recovery between MDTs but not between client/MDT
   * [LU-12631] - Report latency of client operations
   * [LU-12634] - Support for linux kernel version 5.3
   * [LU-12662] - Run pjdfstest POSIX test in review and full testing
   * [LU-12722] - exclude local client mounted on MDS/OSS from recovery
   * [LU-12785] - DOM2: dynamic DoM component size as MDT becomes full
   * [LU-12806] - llapi: have llapi_fd2fid rely on name_to_handle_at() rather than ioctl()
   * [LU-12820] - obd_dirty_transit_pages is always zero and can be removed
   * [LU-12871] - enable changelog garbage collection by default
   * [LU-12899] - Make Lustre build report the required package: kernel-rpm-macros
   * [LU-12905] - sanity/51e can take less with debugging disabled
   * [LU-12910] - osc.*.short_io_bytes parameter cannot be set to the maximum
   * [LU-12929] - Field net_state in struct lnet_net is never used, only set.  It should be discarded
   * [LU-12930] - use schedule_timeout_{un,}interruptible where possible.
   * [LU-12931] - Fix minor errors with cfs_time_seconds usage
   * [LU-12936] - Discard pt_number from lnet_peer_table
   * [LU-12941] - Track aliveness at the peer level
   * [LU-12942] - Optimize check for routing feature flag
   * [LU-12950] - Сhange force_over_512tb lustre mount option to force_over_1024tb for ldiskfs
   * [LU-12966] - print logs when lfsck detected inconsistent_owner
   * [LU-12973] - lustre/doc/.gitignore ignores relevant files
   * [LU-12977] - fix i_mutex for ldiskfs_truncate() in osd_execute_truncate()
   * [LU-12984] - Add --newerXY support for lfs find
   * [LU-12988] - improve mount time on huge ldiskfs filesystem
   * [LU-13004] - Use a single style of IO buf descriptor  - iovec
   * [LU-13005] - Remove LNet event queuing subsystem (lib-eq)
   * [LU-13013] - osd can save credits on mapped blocks
   * [LU-13036] - Reduce memory consumption by lnet.
   * [LU-13056] - lctl llog_catlist, llog_print, etc. should default to MGS device
   * [LU-13059] - kernel update [RHEL7.7 3.10.0-1062.9.1.el7]
   * [LU-13098] - supress connection restore message on idling connections
   * [LU-13110] - kernel update [SLES12 SP4 4.12.14-95.45.1]
   * [LU-13111] - support for SLES12 SP5
   * [LU-13120] - lustre-osd-zfs-mount depends on zfs, kmod-lustre-osd-zfs
   * [LU-13122] - osd-zfs to use 8K blocksize for llog objects
   * [LU-13127] - make Lustre error when hash algo is not available
   * [LU-13130] - sanity-scrub to use total device size with ZFS
   * [LU-13134] - try to use slab allocation for cl_page
   * [LU-13141] - Include block allocator improvements to additional distributions
   * [LU-13151] - Add parent FID to MTIME changelogs
   * [LU-13164] - remove unused LUSTRE_DIRECTIO_FL
   * [LU-13169] - Allow test-framework to repeat subtests
   * [LU-13178] - Update ZFS Version to 0.8.3
   * [LU-13224] - Expose lctl get/set/list param functionality in lustreapi
   * [LU-13226] - SUSE 15 SP2 server ldiskfs
   * [LU-13241] - mount.lustre for large filesystem runs slow debugfs commands
   * [LU-13258] - Bind linux workqueues to specific core
   * [LU-13268] - customize lnet-selftest for performance
   * [LU-13269] - make lnet-selftest.sh more flexible
   * [LU-13272] - customise dom-performance
   * [LU-13274] - Building againt lustreapi using -std=c99
   * [LU-13275] - enhance racer with various FOFB scenarios
   * [LU-13280] - customize run_mdtest() for several mount points
   * [LU-13281] - ha.sh improvements: different layouts  for clients directories
   * [LU-13283] - ha.sh improvement: add racer to nonmpi load
   * [LU-13297] - parallel-scale enhancement
   * [LU-13312] - Optimized RA for stride read under memory pressure
   * [LU-13347] - kernel update RHEL 7.8 
   * [LU-13366] - SEL addons
   * [LU-13369] - kernel update [RHEL7.7 3.10.0-1062.18.1.el7]
   * [LU-13383] - lazy atime update on OST
   * [LU-13388] - racer crash: general protection fault: 0000 RIP: lod_obj_for_each_stripe+0xf7/0x2d0 [lod]
   * [LU-13394] - avoid sending OST RPC for project setting
   * [LU-13405] - kernel update [SLES15 SP1 4.12.14-197.37.1]
   * [LU-13412] - fix read if readahead window smaller than rpc size
   * [LU-13421] - kernel update [RHEL8.1 4.18.0-147.8.1.el8_1]
   * [LU-13445] - enhance ha.sh to support different users
   * [LU-13488] - Support RHEL 8.2
   * [LU-13501] - Disable health on single-rail deployments
   * [LU-13502] - Provide ability to fine-tune response tracking
   * [LU-13503] - allow setting larger max_mod_rpcs_in_flight dynamically
   * [LU-13509] - Improve ptlrpc_register_bulk() behavior
   * [LU-13510] - Allow control over LND timeouts independent of lnet_transaction_timeout and retry_count
   * [LU-13533] - Disable lazy_itable_init
   * [LU-13556] - kernel update [RHEL7.8 3.10.0-1127.8.2.el7]
   * [LU-13571] - Refine which network errors result in LNet Health activity
   * [LU-13585] - add mustfail check to ha.sh
   * [LU-13590] - support for RHEL 7.9
   * [LU-13595] - add option to lustre_rmmod to dump debug logs
   * [LU-13597] - add processing time/latency, IO sizes, stats to job_stats
   * [LU-13600] - limit number of RPCs in flight during recovery
   * [LU-13651] - Conditionally skip finding compatible HSM requests
   * [LU-13657] - kernel update [RHEL8.2 4.18.0-193.6.3.el8_2]
   * [LU-13676] - script to show unique backtraces from crash utility
   * [LU-13699] - Improve: mdt_identity.c Error handling and Error message reporting
   * [LU-13723] - Use ERR_CAST() instead of ERR_PTR(PTR_ERR()) 
   * [LU-13734] - Allow duplicate nets in kernel ip2nets syntax
   * [LU-13735] - Loosen restrictions on LNet Health parameter settings
   * [LU-13740] - Ubuntu 20.04 LTS release
   * [LU-13761] - MOFED 5.1 support
   * [LU-13782] - LNet Routers should monitor the ni_fatal flag to inform peers of changes to route status
   * [LU-13819] - Update ZFS Version to 0.8.4
   * [LU-13827] - ofd_access_log_reader to print top hot files
   * [LU-13837] - Introduce constant for LNET_NIDNET(LNET_NID_ANY)
   * [LU-13839] - support for RHEL 8.3
   * [LU-13919] - kernel update [RHEL7.8 3.10.0-1127.19.1.el7]
   * [LU-13921] - change output format for ofd_access_log_reader
   * [LU-13944] - t-f: add ability to set clients parameters
   * [LU-13961] - kernel update [RHEL8.2 4.18.0-193.19.1.el8_2]
   * [LU-13969] - Updates to lustre-release yaml.sh
   * [LU-13977] - sanity test_255a uses integer comparison for non-integers
   * [LU-14020] - Update SUSE SP2 ldiskfs
   * [LU-14029] - kernel update [SLES15 SP2 5.3.18-24.24.1]
   * [LU-14043] - lfs mirror extend need not use O_DIRECT on source
   * [LU-14052] - ofd_access_log_reader support for multiple MDTs
   * [LU-14054] - add option to ofd_access_log_reader to exit on close of last log
   * [LU-14071] - add OST pool quota options for lfs-quota.1 and lfs-setquota.1
   * [LU-14075] - kernel update [RHEL8.2 4.18.0-193.28.1.el8_2]
   * [LU-14077] - kernel update [SLES15 SP1 4.12.14-197.64.1]
   * [LU-14129] - kernel update [RHEL7.9 3.10.0-1160.6.1.el7]
   * [LU-14134] - reduce credits for new writing potentially
   * [LU-14140] - osd_idc_find_and_init() should dump details
   * [LU-14154] - ldiskfs doesn't builld with gcc10
   * [LU-14170] - Update ZFS version to 2.0.0
   * [LU-14194] - add mdc/mdd related checksum parameter files
   * [LU-14218] - kernel update [SLES15 SP1 4.12.14-197.75.1]
   * [LU-14265] - Allow "lfs mkdir" without arguments
   * [LU-14334] - update lnet/ChangeLog with current information

• • Technical task

   * [LU-4423] - Tracking of patches from upstream kernel to Lustre client
   * [LU-10664] - DoM: make DoM lock enqueue non-blocking
   * [LU-11300] - LNet: Router Aliveness and Health
   * [LU-12923] - replace CLASSERT() with BUILD_BUG_ON()
   * [LU-13415] - DoM: keep optimizations even when OST stripe is initialized