Whamcloud - gitweb
Patrick Farrell [Wed, 8 May 2024 14:36:55 +0000 (10:36 -0400)]
LU-17831 llite: remove refcount assert
This refcount assert costs us a few % in page freeing and
it comes immediately after a dec_and_test.
Signed-off-by: Patrick Farrell <patrick.farrell@oracle.com>
Change-Id: I8de3369cbbaf484e1e0fce27ae3e62cad1ae5282
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/55054
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
Reviewed-by: Arshad Hussain <arshad.hussain@aeoncomputing.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Patrick Farrell [Wed, 8 May 2024 14:36:33 +0000 (10:36 -0400)]
LU-17831 llite: remove PageLocked assert
This pagelocked assert touches the spinlock and costs us
about 5% of our performance when dropping pages.
Signed-off-by: Patrick Farrell <patrick.farrell@oracle.com>
Change-Id: Ie67f45fc0131d3fa94b70e3734ade2139c481301
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/55053
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
Reviewed-by: Arshad Hussain <arshad.hussain@aeoncomputing.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Sebastien Buisson [Wed, 20 Mar 2024 14:45:41 +0000 (15:45 +0100)]
LU-17431 nodemap: allow handling a nidrange for a dynamic nm
Adding or deleting a nid range for a dynamic nodemap, by sending the
appropriate ioctl to the MDS or OSS device.
On kernel side, we need to allow handling a range for a nodemap even
if the nodemap config file is not accessible (i.e. we are not on the
MGS). This dynamic nodemap is not stored on disk.
Signed-off-by: Sebastien Buisson <sbuisson@ddn.com>
Change-Id: I34d6bb720cb700e23a3567a47def71c2b1ca7343
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/54508
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
Reviewed-by: James Simmons <jsimmons@infradead.org>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Sebastien Buisson [Wed, 20 Mar 2024 09:05:22 +0000 (10:05 +0100)]
LU-17431 nodemap: allow add/del of a dynamic nodemap
Adding a dynamic nodemap requires specifying the '-d' option to
'lctl nodemap_add'. This in turn sends the appropriate ioctl to
the MDS or OSS device.
Deleting a dynamic nodemap does not require this flag, as we can
figure out if the nodemap to be deleted is dynamic or not.
On kernel side, we need to allow handling a nodemap even if the
nodemap config file is not accessible (i.e. we are not on the MGS).
This dynamic nodemap is not stored on disk.
And we prevent modifications of static, on-disk nodemaps on non-MGS
servers. On these servers it is only possible to modify dynamic
nodemaps.
Signed-off-by: Sebastien Buisson <sbuisson@ddn.com>
Change-Id: I10019ad131825c3db520689e01b4c931d90c1c89
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/54507
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
Reviewed-by: James Simmons <jsimmons@infradead.org>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Lei Feng [Thu, 6 Jan 2022 01:34:15 +0000 (20:34 -0500)]
LU-10499 pcc: add some statistics data
Add statictics of the number and total size of pcc attached files
and pcc hit files.
EX-bug-id: EX-4433
Change-Id: Ib0e429c636298d4c6ff06d84a416073895b86184
Signed-off-by: Lei Feng <flei@whamcloud.com>
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/54458
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Reviewed-by: James Simmons <jsimmons@infradead.org>
Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
Qian Yingjin [Wed, 20 Oct 2021 04:07:14 +0000 (12:07 +0800)]
LU-10499 pcc: command to remove PCC mirror component
This patch adds a command "lfs pcc delete $FILE" to delete the
PCC foreign mirror layout component.
Describe fields in lfs-pcc-state.1 and lfs-pcc-delete.1 page.
EX-bug-id: EX-4055
Signed-off-by: Qian Yingjin <qian@ddn.com>
Signed-off-by: Andreas Dilger <adilger@whamcloud.com>
Change-Id: I3f56fb8134bd1e7673ef8e04dff9b8482f0e32c3
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/54457
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: James Simmons <jsimmons@infradead.org>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Qian Yingjin [Wed, 29 Sep 2021 03:35:30 +0000 (11:35 +0800)]
LU-10499 pcc: Check if PCC copy is unlinked for state output
In this patch, it adds support for the command "lfs pcc state" to
check whether the PCC copy is in the local client cache or
unlinked improperly.
Do not print an error message if "lfs pcc detach" tries to detach
a file that is already removed from the cache. This might happen
for a wide variety of reasons (external cache cleanup process, etc).
EX-bug-id: EX-3825
Test-Parameters: testlist=sanity-pcc env=ONLY=48
Signed-off-by: Qian Yingjin <qian@ddn.com>
Change-Id: Ic50c901df78adfaf5b56990120f832e5d74a117c
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/54451
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
Reviewed-by: Li Xi <lixi@ddn.com>
Reviewed-by: James Simmons <jsimmons@infradead.org>
Patrick Farrell [Tue, 12 Nov 2024 22:35:49 +0000 (17:35 -0500)]
LU-18442 mdc: remove PageChecked usage in mdc
PageChecked is set in mdc_readpage, and it's not clear why.
This was added in an early merge of a bunch of code, but
was never checked anywhere.
It's unclear what this ever did, but it's not doing
anything now. Remove.
Signed-off-by: Patrick Farrell <pfarrell@whamcloud.com>
Change-Id: I59d40cc5cb9b437182db04b468f7dc222c275424
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/52504
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
Reviewed-by: Mikhail Pershin <mpershin@whamcloud.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Shaun Tancheff [Sat, 23 Sep 2023 07:02:30 +0000 (02:02 -0500)]
LU-17139 utils: l_getidentity remove 'files' => 'lustre' alias
Fully remove the old 'files' alias for 'lustre' and deprecate
'nss_files' alias for 'files' now that it can be used directly.
Signed-off-by: Shaun Tancheff <shaun.tancheff@hpe.com>
Change-Id: Ied7cf87e9be8d474f9b65a9b5b14870578806151
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/52487
Reviewed-by: corey tesdahl <corey.tesdahl@hpe.com>
Reviewed-by: Petros Koutoupis <petros.koutoupis@hpe.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Etienne AUJAMES [Wed, 2 Aug 2023 09:27:41 +0000 (11:27 +0200)]
LU-16235 hsm: get a valid cookie for RAoLU request
Add a way to get a valid cookie when nobody initializes
cdt_last_cookie.
RAoLU policy is allowed to queue a remove request with the
coordinator stopped. In that cases cdt_last_cookie can not be yet
initialize and the remove request can be queued with a conflicting
cookie.
This patch adds cdt_update_last_cookie() that reverses process the hsm
llog and stops at the first non-cancel action to determine the last
cookie.
Add the regression test sanity-hsm 26e.
Test-Parameters: testlist=sanity-hsm
Test-Parameters: testlist=sanity-hsm
Test-Parameters: testlist=sanity-hsm
Test-Parameters: testlist=sanity-hsm env=ONLY=26e,ONLY_REPEAT=30
Signed-off-by: Etienne AUJAMES <eaujames@ddn.com>
Signed-off-by: Nikitas Angelinas <nikitas.angelinas@hpe.com>
Change-Id: I6468a24b95fcb8768e12f40edfcea3ce8407281f
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/51850
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Qian Yingjin [Thu, 6 Jul 2023 13:21:40 +0000 (09:21 -0400)]
LU-14361 statahead: add tunable for fname pattern detection
This patch adds two tunable parameters for the detection of the
fname pattern statahead:
- llite.*.statahead_fname_predict_hit: when the naming of stat()
files under a directory follows a certain name rule roughly more
than this value, the directory is considered to meet the first
requirement for statahead. For an example, file naming rule is
mdtest.$rank.$i, the suffix of the stat() dentry name is number
and do stat() for dentries with name ending with number more
then this parameter.
- llite.*.statahead_fname_match_hit: After meet the first
requirement, then the naming of stat() files under a directory
continuously satisfies a certain name rule strictly more than
this value, it will start a statahead thread to do attribute
prefetching under the directory.
This patch also fixes the following panic:
IP: _atomic_dec_and_lock+0xc/0x70
->ll_sax_put [lustre]
->ll_statahead_thread [lustre]
->kthread
The reason is that the @lli_sax is set with NULL by wrong
statahead context (sax) put.
This patch also fixes the possible deadlock between hardlink and
batch stat-ahead operations.
Fix the test failure on lustre-rsync-test/test_6.
It must wait all inuse inodes for statahead to release during
umount.
Otherwsie, it may cause the following panic:
BUG: Dentry
0000000033ca4f3e{i=
280001b840002c7,n=l4} still in use
(1) [unmount of lustre lustre]
RIP: 0010:umount_check.cold.52+0x2f/0x3b
d_walk+0xe7/0x290
do_one_tree+0x20/0x40
shrink_dcache_for_umount+0x28/0x90
generic_shutdown_super+0x1a/0x110
kill_anon_super+0x14/0x30
deactivate_locked_super+0x34/0x70
cleanup_mnt+0x3b/0x70
RIP: 0010:ll_prep_md_op_data+0x73/0x870 [lustre]
sa_prep_data+0xde/0x350 [lustre]
sa_statahead+0x3b9/0xd20 [lustre]
ll_statahead_thread+0x1507/0x21f0 [lustre]
kthread+0x134/0x150
Test-Parameters: clientdistro=el8.10 testlist=sanity
Test-Parameters: clientdistro=el8.10 testlist=sanity
Test-Parameters: clientdistro=el8.10 testlist=sanity
Test-Parameters: clientdistro=el8.10 testlist=sanity
Test-Parameters: clientdistro=el8.10 testlist=sanity
Signed-off-by: Qian Yingjin <qian@ddn.com>
Change-Id: I42d9478e796918d9f2498ab64cf7c20b61334144
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/51592
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Lai Siyao <lai.siyao@whamcloud.com>
Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
James Simmons [Sun, 18 Aug 2024 00:18:18 +0000 (20:18 -0400)]
LU-11850 obd: use netlink to get lustre stats
This adds the ability to collect performance metrics from lustre
in another way then from proc / debugfs files. The move to debugfs
has limited the scope of access to only root. Additionally there
is an expensive cost accessing many virtual file system files to
collect that data. Netlink will scale much better in this case as
well as offer a much more flexiable API.
The new ldebugfs_stats_alloc() replaces lprocfs_stats_alloc() and
registers the stats to be accessible BOTH throught debugfs AND
through Netlink. The new global "lstats_list" contains a list of
all registered sets of statistics, so it mirrors a subset of
debugfs. Netlink access can report on any statistics registered in
lstats_list.
Test-Parameters: trivial
Change-Id: If2d662baa62348fe6f0dd5c8d77344650c2a27d8
Signed-off-by: James Simmons <jsimmons@infradead.org>
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/34256
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Shaun Tancheff <shaun.tancheff@hpe.com>
Reviewed-by: Neil Brown <neilb@suse.de>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Elena Gryaznova [Fri, 27 Sep 2024 12:13:27 +0000 (15:13 +0300)]
LU-11073 tests: enable DNE in recovery-mds-scale
In recovery-mds-scale.sh, recovery-random-scale.sh,
recovery-double-scale.sh add environmental variables
RECOVERY_SCALE_ENABLE_REMOTE_DIRS,
RECOVERY_SCALE_ENABLE_STRIPED_DIRS
to enable or disable the corresponding operations.
RECOVERY_SCALE_ENABLE_REMOTE_DIRS defaults to true if
the MDS version is at least 2.5, while
RECOVERY_SCALE_ENABLE_STRIPED_DIRS defaults to true if
the MDS version is at least 2.8.
Patch also:
simplifies the run_*.sh logic: it removes the never
used BREAK_ON_ERROR and useless CONTINUE parameters;
adds the possibility to set the directory striping
via client_load_SETSTRIPEPARAMS;
ignores failures caused by ENOSPC error: instead of
adding the complex free space calculation logic
for various layouts let's allow the loads to fail with
ENOSPC. This does not affect fofb testing the recovery
scale tests are for.
forces lustre-rsync-test test_4 to run iozone on not
striped directory with offset=0
Test-Parameters: trivial env=SLOW=no,FAILURE_MODE=HARD,RACER_ENABLE_REMOTE_DIRS=true,RACER_ENABLE_STRIPED_DIRS=true clientcount=4 mdtcount=1 mdscount=2 osscount=2 austeroptions=-R failover=true iscsi=1 testlist=recovery-mds-scale,recovery-oss-scale,recovery-random-scale,lustre-rsync-test,racer
Signed-off-by: Elena Gryaznova <elena.gryaznova@hpe.com>
Signed-off-by: Andriy Skulysh <andriy.skulysh@hpe.com>
HPE-bug-id: LUS-5973, LUS-6386, LUS-7540, LUS-9399
Reviewed-by: Vladimir Saveliev <vlaidimir.saveliev@hpe.com>
Reviewed-by: Alexander Boyko <alexander.boyko@hpe.com>
Reviewed-by: Alexander Zarochentsev <alexander.zarochentsev@hpe.com>
Change-Id: Icd42ae9b4e5ac403ba76a9e3909616977dbd6a72
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/32626
Tested-by: Maloo <maloo@whamcloud.com>
Tested-by: jenkins <devops@whamcloud.com>
Reviewed-by: Shaun Tancheff <shaun.tancheff@hpe.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Ronnie Sahlberg [Wed, 4 Sep 2024 04:48:48 +0000 (00:48 -0400)]
LU-11610 target: Fix correct return value in out_handler.c
out_handler.c has a pattern:
if (IS_ERR(x) || expression) {
...
RETURN(PTR_ERR(x))
}
which if x is a valid pointer but the expression that represents an
error would cause the function to return PTR_ERR(x) for a valid pointer.
Fix this by either returning PTR_ERR(x) or -EPROTO depending on what the error
is.
Signed-off-by: Ronnie Sahlberg <rsahlberg@whamcloud.com>
Change-Id: Ic9f3427be31a9ba84b46f349ba90cc6aea379845
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56260
Reviewed-by: Li Dongyang <dongyangli@ddn.com>
Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
Reviewed-by: Timothy Day <timday@amazon.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Shaun Tancheff [Thu, 21 Nov 2024 01:55:54 +0000 (08:55 +0700)]
LU-18471 ldiskfs: large directory causes htree corruption
When creating a lot of files in a single directory, it can
get corrupted because of a typo in ext4-kill-dx-root.patch.
Lustre-change: https://review.whamcloud.com/46526
Lustre-commit:
ea3ee9337f9bcd42360e4523f1e34bcd04d3bf41
Test-Parameters: trivial
HPE-bug-id: LUS-12617
Fixes:
8da23f070c ("LU-15544 ldiskfs: SUSE 15 SP4 kernel 5.14.21 SUSE")
Fixes:
fc87b01f96 ("LU-12477 ldiskfs: remove obsolete ext4 patches")
Fixes:
89075044b3 ("LU-12477 ldiskfs: drop SUSE kernel 4.4 and earlier")
Signed-off-by: Shaun Tancheff <shaun.tancheff@hpe.com>
Change-Id: Iacbaf9840db76ea7e2e017835a14b476ca9be391
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/57094
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
Reviewed-by: Alex Zhuravlev <bzzz@whamcloud.com>
Reviewed-by: Artem Blagodarenko <ablagodarenko@ddn.com>
Reviewed-by: Andrew Perepechko <andrew.perepechko@hpe.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Sergey Cheremencev [Fri, 17 May 2024 18:25:28 +0000 (21:25 +0300)]
LU-15913 tests: fix set_params_xxx
Move argument number check from set_params_(clients|mdts|osts)
to set_params_nodes. Without that fix these functions
didn't set (MDS|OSS|CLIENT)_LCTL_SETPARAM_PARAM when
the number of nodes was < 2.
Test-Parameters: trivial
Fixes:
9a1d68f9b8 ("LU-15913 tests: add rename stress test via racer")
Signed-off-by: Sergey Cheremencev <scherementsev@ddn.com>
Change-Id: I0733109c53cdea435b4461da7ac44e54fa49498c
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/55143
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
Reviewed-by: Alex Deiter <adeiter@ddn.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Alex Zhuravlev [Sun, 10 Nov 2024 07:17:37 +0000 (10:17 +0300)]
LU-18435 lod: recover layout generation from replay
The offset of the layout generation is different between struct
lov_mds_md_v1/v3.lmm_layout_gen and lov_comp_md.lcm_layout_gen.
When checking/setting layout gen, we must use layout-specific field.
Otherwise layout generation can be set to 0 (or other random value)
after replay and client can't apply new layout during later update.
Fixes:
13557aa86904 ("LU-15300 mdt: refresh LOVEA with LL granted")
Signed-off-by: Alex Zhuravlev <bzzz@whamcloud.com>
Change-Id: I5e4a63cd097d157317e0e8d1a0fca4a46817d118
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56950
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Reviewed-by: Qian Yingjin <qian@ddn.com>
Reviewed-by: Alexey Lyashkov <alexey.lyashkov@hpe.com>
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Etienne AUJAMES [Thu, 10 Oct 2024 12:06:10 +0000 (14:06 +0200)]
LU-15535 lmv: fix lmv_stripe_object_dump() CDEBUG
"refs" and "magic" are reversed in CDEBUG() arguments of
lmv_stripe_object_dump().
Test-Parameters: trivial
Fixes:
3ebc8e0528e3 ("LU-15535 llite: deadlock on lli_lsm_sem")
Signed-off-by: Etienne AUJAMES <eaujames@ddn.com>
Change-Id: Ibb4dc6ea38b506f7262da92699cf5ed97a362f3d
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56647
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
Reviewed-by: Qian Yingjin <qian@ddn.com>
Reviewed-by: Lai Siyao <lai.siyao@whamcloud.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Etienne AUJAMES [Fri, 27 Sep 2024 14:50:15 +0000 (16:50 +0200)]
LU-18260 o2iblnd: fix race between REJ vs kiblnd_connd
This patch fixes a possible race between CM_EVENT_REJECTED and
kiblnd_connd().
kiblnd_connd() set connection state to IBLND_CONN_DISCONNECTED
before removing the QP. So if CM_EVENT_REJECTED is received in this
time windows, it will cause the following crash:
Workqueue: ib_cm cm_work_handler [ib_cm]
all Trace:
<TASK>
dump_stack_lvl+0x34/0x48
panic+0x100/0x2d2
lbug_with_loc.cold+0x18/0x18 [libcfs]
kiblnd_cm_callback+0x108d/0x10b0 [ko2iblnd]
cma_cm_event_handler+0x1e/0xb0 [rdma_cm]
cma_ib_handler+0x8d/0x2e0 [rdma_cm]
cm_process_work+0x22/0x190 [ib_cm]
cm_rej_handler+0xdf/0x260 [ib_cm]
cm_work_handler+0x47f/0x4d0 [ib_cm]
process_one_work+0x1e8/0x390
worker_thread+0x53/0x3d0
kthread+0x124/0x150
ret_from_fork+0x1f/0x30
</TASK>
Test-Parameters: trivial testlist=sanity-lnet
Fixes: 0b8c18d ("LU-17480 o2iblnd: add a timeout for rdma_connect")
Signed-off-by: Etienne AUJAMES <eaujames@ddn.com>
Change-Id: I2d04433eb51e1a6862b788a89e127d8abb24b8a9
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56518
Reviewed-by: Frank Sehr <fsehr@whamcloud.com>
Reviewed-by: Serguei Smirnov <ssmirnov@whamcloud.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Andriy Skulysh [Tue, 5 Feb 2019 13:37:48 +0000 (15:37 +0200)]
LU-2525 ldlm: add asynchronous flocks
Add support of asynchronous flocks.
They are used only by Linux nfsd for now.
HPE-bug-id: LUS-3210, LUS-7034,LUS-7031,LUS-8832, LUS-8313
HPE-bug-id: LUS-8592
Change-Id: Iefafaf014fd06d569dc5d1dd22ebb3518d04e99a
Reviewed-by: Vitaly Fertman <c17818@cray.com>
Reviewed-by: Alexander Boyko <c17825@cray.com>
Signed-off-by: Andriy Skulysh <andriy.skulysh@hpe.com>
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/4889
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Alexander Boyko <alexander.boyko@hpe.com>
Reviewed-by: Alexey Lyashkov <alexey.lyashkov@hpe.com>
Reviewed-by: Vitaly Fertman <vitaly.fertman@hpe.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Emoly Liu [Sun, 3 Nov 2024 10:31:12 +0000 (18:31 +0800)]
LU-18390 tests: dump debug logs during module unload correctly
If DEBUG is enabled, debug logs will be dumped to:
- $TMP/debug by default if DEBUG_RMMOD is unset. And if memory
leak is found, $TMP/debug will be renamed to $TMP/debug-leak.xxx;
- $DEBUG_RMMOD if $DEBUG_RMMOD is a full path name;
- $TMP/DEBUG_RMMOD if $DEBUG_RMMOD is a filename;
- standard output if DEBUG_RMMOD=- .
Memory leak in conf-sanity.sh test_29 in interop testing(LU-17962)
is used to verify this patch.
Test-Parameters: env=DEBUG=true,IGNORE_LEAK=yes,ONLY=29 \
testlist=conf-sanity serverversion=2.15 clientdistro=el8.10 \
serverdistro=el8.10 mdscount=2 mdtcount=4 ostcount=8
Test-Parameters: env=DEBUG_RMMOD=lu18390,IGNORE_LEAK=yes,ONLY=29 \
testlist=conf-sanity serverversion=2.15 clientdistro=el8.10 \
serverdistro=el8.10 mdscount=2 mdtcount=4 ostcount=8
Test-Parameters:trivial
Fixes:
255102e84e ("LU-16384 tests: dump lustre log if DEBUG_RMMOD set")
Signed-off-by: Emoly Liu <emoly@whamcloud.com>
Change-Id: I23a584541d0f9b313cf00e56f63dc4ac356c3cbc
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56767
Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
Reviewed-by: Alex Zhuravlev <bzzz@whamcloud.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Mikhail Pershin [Tue, 8 Oct 2024 13:10:37 +0000 (16:10 +0300)]
LU-18352 mgc: explicitly create sptlrpc local copy
Sptlrpc config has single instance per MGC and is shared
by targets, hence it is processed only once and local
copy of it is created also only during first processing
for the first target doing that. All other targets just
find config in memory.
Therefore a local copy creation for other targets need
to be done explicitly when already processed config
is found.
Patch introduces mgc_get_local_copy() which does
just llog copy from MGS if possible for a target
finding already processed sptlrpc config
Test-Parameters: testlist=sanity-sec env=ONLY=70,SHARED_KEY=true
Signed-off-by: Mikhail Pershin <mpershin@whamcloud.com>
Change-Id: I922f92a950b9a07172f36f42b94da854c7702a80
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56609
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Sebastien Buisson <sbuisson@ddn.com>
Reviewed-by: Aurelien Degremont <adegremont@nvidia.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Jian Yu [Tue, 8 Oct 2024 19:42:49 +0000 (12:42 -0700)]
LU-18338 build: fix Ubuntu kernel signature check
The Ubuntu kernel signature was checked by searching for
CONFIG_VERSION_SIGNATURE from autoconf.h. However, in
Ubuntu kernel 6.11.0-061100-generic, CONFIG_VERSION_SIGNATURE
is not defined. We can search for UTS_UBUNTU_RELEASE_ABI
from utsrelease.h.
Test-Parameters: trivial mdtcount=4 mdscount=2 \
clientdistro=ubuntu2404 testlist=sanity
Test-Parameters: trivial mdtcount=4 mdscount=2 \
env=SANITY_EXCEPT="255c" \
clientdistro=ubuntu2204 testlist=sanity
Change-Id: I62e74ef936bbbf4e85130965cfff35aa7aa3be5e
Signed-off-by: Jian Yu <yujian@whamcloud.com>
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56613
Reviewed-by: James Simmons <jsimmons@infradead.org>
Reviewed-by: Shaun Tancheff <shaun.tancheff@hpe.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Alex Zhuravlev [Sun, 6 Oct 2024 14:47:02 +0000 (17:47 +0300)]
LU-18334 obdclass: wait for RCU completion
in lu_kmem_fini(), otherwise otherwise those RCU callbacks doing
kmem_cache_free() can race with kmem_cache_destroy():
kmem_cache_destroy echo_object_kmem: Slab cache still has objects
WARNING: CPU: 1 PID: 7991 at mm/slab_common.c:523 kmem_cache_destroy
Signed-off-by: Alex Zhuravlev <bzzz@whamcloud.com>
Change-Id: I21dfc034b9fc9368bf22d269d6986297a6812a5c
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56593
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Yang Sheng <ys@whamcloud.com>
Reviewed-by: James Simmons <jsimmons@infradead.org>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Arshad Hussain [Mon, 30 Sep 2024 07:35:45 +0000 (03:35 -0400)]
LU-18277 tests: Add missing env variable for acceptance-small
Although LUSTRE and NAME is defined once auster is
called via test-framwork.sh inclusion. It is still
undefined at this point making running standalone
acceptance-small.sh to fail. This patch defines
missing env variable LUSTRE and NAME when running
standalone accpetance-small.sh
Test-Parameters: trivial
Signed-off-by: Arshad Hussain <arshad.hussain@aeoncomputing.com>
Change-Id: Ib540f4ce7d56f6205f4d9e43bbc3b9e6db94511b
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56538
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Alex Deiter <adeiter@ddn.com>
Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Sebastien Buisson [Fri, 27 Sep 2024 07:48:44 +0000 (09:48 +0200)]
LU-18256 gss: deprecate insecure enctypes
A number of encryption types declared in the GSS code are deprecated
for security reasons, and should not be used. So remove support for
them in the Lustre code:
- des-cbc-crc
- des-cbc-md4
- des-cbc-md5
- des-cbc-raw
- des-hmac-sha1
- des3-cbc-sha
- des3-cbc-raw
- des3-cbc-sha1
- arcfour-hmac
- arcfour-hmac-exp
Test-Parameters: trivial
Test-Parameters: testgroup=review-dne-selinux-ssk-part-1
Test-Parameters: testgroup=review-dne-selinux-ssk-part-2
Test-Parameters: kerberos=true testlist=sanity-krb5
Signed-off-by: Sebastien Buisson <sbuisson@ddn.com>
Change-Id: Ic8dd2470339323be88a416796c8d420ecd2f55e4
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56512
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
Reviewed-by: Aurelien Degremont <adegremont@nvidia.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Shaun Tancheff [Thu, 26 Sep 2024 10:46:42 +0000 (17:46 +0700)]
LU-18272 build: remove Summary line from osd-zfs
Resolve spurious warning:
warning: line 390: second Summary
when building src rpm:
Test-Parameters: trivial
HPE-bug-id: LUS-12538
Signed-off-by: Shaun Tancheff <shaun.tancheff@hpe.com>
Change-Id: I6aa591aae3ae4dc07a36740e12ef3520cea035ef
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56506
Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
Reviewed-by: corey tesdahl <corey.tesdahl@hpe.com>
Reviewed-by: Petros Koutoupis <petros.koutoupis@hpe.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Sebastien Buisson [Mon, 23 Sep 2024 08:31:07 +0000 (10:31 +0200)]
LU-18256 gss: support SHA2 enctypes
Introduce support for ENCTYPE_AES128_CTS_HMAC_SHA256_128 and
ENCTYPE_AES256_CTS_HMAC_SHA384_192 encryption types that are used by
GSS code for authentication context.
Test-Parameters: trivial
Test-Parameters: testgroup=review-dne-selinux-ssk-part-1
Test-Parameters: testgroup=review-dne-selinux-ssk-part-2
Test-Parameters: kerberos=true testlist=sanity-krb5
Signed-off-by: Sebastien Buisson <sbuisson@ddn.com>
Change-Id: I42ab758b42b24c64647cd771887a2fd26bc55394
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56489
Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
Reviewed-by: Aurelien Degremont <adegremont@nvidia.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Qian Yingjin [Wed, 11 Sep 2024 03:06:40 +0000 (11:06 +0800)]
LU-17190 osc: account DIO in flight using server-side locking
Add accounting for DIO (using server-side DLM extent locking) in
flight.
We can see from osc.*.rpc_stats to judge whether all I/O RPC slots,
which are limit by @max_rpcs_in_flight, are used out by DIOs.
Change-Id: I3d2c6e607d19037bb399c2bceb50e64826263469
Signed-off-by: Qian Yingjin <qian@ddn.com>
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56325
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Shaun Tancheff <shaun.tancheff@hpe.com>
Reviewed-by: Patrick Farrell <pfarrell@whamcloud.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Shaun Tancheff [Tue, 10 Sep 2024 08:58:56 +0000 (15:58 +0700)]
LU-16350 ldiskfs: update for kernel 6.11
Update ext4-kill-dx-root.patch and ext4-lookup-dotdot.patch
for kernel 6.11.
These updates are also applicable to 6.10 stable, tested with
6.10.6 stable.
HPE-bug-id: LUS-11376
Test-Parameters: trivial
Signed-off-by: Shaun Tancheff <shaun.tancheff@hpe.com>
Change-Id: I94f95eeff65b80f879a8b34aea05dc5fa289aa73
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56284
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
Reviewed-by: Li Dongyang <dongyangli@ddn.com>
Reviewed-by: Jian Yu <yujian@whamcloud.com>
Shaun Tancheff [Sat, 21 Sep 2024 04:01:42 +0000 (11:01 +0700)]
LU-18190 build: compatibility updates for kernel 6.11
Linux commit v6.10-12269-g78eb4ea25cd5
sysctl: treewide: constify the ctl_table argument of proc_handlers
Constify ctl_table and cast away const for older kernels
Linux commit v5.15-rc3-13-g9257e1567738
mm/filemap: Add folio_index(), folio_file_page() and
folio_contains()
Linux commit v6.10-rc6-27-g05b0c7edad9b
mm: drop page_index and simplify folio_index
Removed page_index() in favor of folio_index() provide a wrapper
folio_index_page() to call the correct function with a page.
Linux commit v6.1-rc4-186-gcb67f4282bf9
mm,thp,rmap: simplify compound page mapcount handling
Adds folio_mapcount()
Linux commit v6.10-rc6-100-gcdd9a571b7d8
fs/proc: move page_mapcount() to fs/proc/internal.h
Removed page_mapcount() in favor of folio_mapcount() provide
folio_mapcount_page() to call page_mapcount() or calculate
mapcount as done by folio_precise_page_mapcount()
Linux commit v6.10-rc3-19-ge9f5f44ad372
block: remove the blk_integrity_profile structure
Linux commit v6.10-rc3-25-g9f4aa46f2a74
block: invert the BLK_INTEGRITY_{GENERATE,VERIFY} flags
Invert the checks for BLK_INTEGRITY_NO{GENERATE,VERIFY} when
BLK_INTEGRITY_NOVERIFY is present and remove checks that
require bi->profile->{verify_fn,generate_fn}
Also resolve a gcc-14 issue with -Werror=stringop-truncation
in lustre/utils/obd.c
HPE-bug-id: LUS-12519
Test-Parameters: trivial
Signed-off-by: Shaun Tancheff <shaun.tancheff@hpe.com>
Change-Id: Ifda4da9716108129bb59634612940d61abe69aa2
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56283
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
Reviewed-by: Jian Yu <yujian@whamcloud.com>
Shaun Tancheff [Fri, 6 Sep 2024 05:51:12 +0000 (12:51 +0700)]
LU-18213 o2ib: dec connection count before peer
BUG: KFENCE: use-after-free write in \
kiblnd_destroy_conn+0x356/0x660 [ko2iblnd]
In kiblnd_destroy_conn() calling kiblnd_peer_decref()
could result in freeing the peer_ni.
Drop the connection counters before calling
kiblnd_peer_decref() to avoid use after free
HPE-bug-id: LUS-12513
Signed-off-by: Shaun Tancheff <shaun.tancheff@hpe.com>
Change-Id: I85fcfc399ca38e9b85d9eff72314f0363e2a0666
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56275
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Chris Horn <chris.horn@hpe.com>
Reviewed-by: Petros Koutoupis <petros.koutoupis@hpe.com>
Reviewed-by: James Simmons <jsimmons@infradead.org>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Qian Yingjin [Mon, 2 Sep 2024 08:48:12 +0000 (16:48 +0800)]
LU-18192 nrs: move nid/uid/gid/opcode/jobid into generic key
A more generic data struct ("nrs_tbf_key") is defined to store the
nid/uid/gid/opcode/jobid information of a TBF bucket (data struct
nrs_tbc_client).
Remove the old tc_key for the generic NRS TBF type. And replace
with this new data struct (nrs_tbf_key) instead.
Signed-off-by: Qian Yingjin <qian@ddn.com>
Change-Id: I0f62f02eb358d1ef697bbacd2e8225956deff0ec
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56227
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
Reviewed-by: Etienne AUJAMES <eaujames@ddn.com>
Reviewed-by: James Simmons <jsimmons@infradead.org>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Shaun Tancheff [Mon, 26 Aug 2024 04:02:17 +0000 (11:02 +0700)]
LU-17000 lnet: lnet_inet_enumerate krealloc and kfree
CoverityID: 442369 ("Memory - corruptions, double free")
On realloc() failure free 'ifaces' which may already differ from
*dev_list due to a previous realloc(). Also nalloc is now zero.
CoverityID: 442378 ("Resource Leak")
Ensure lnet_inet_enumerate() ifaces are freed.
Signed-off-by: Shaun Tancheff <shaun.tancheff@hpe.com>
Change-Id: I2d2762d86fcf070387b100115ad3a50bd2b2840b
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56150
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Frank Sehr <fsehr@whamcloud.com>
Reviewed-by: James Simmons <jsimmons@infradead.org>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Reviewed-by: Chris Horn <chris.horn@hpe.com>
Shaun Tancheff [Tue, 27 Aug 2024 02:59:35 +0000 (09:59 +0700)]
LU-17000 utils: Coverity leaks and races
CoverityID: 442365 ("time-of-check, time-of-use")
CoverityID: 442366 ("time-of-check, time-of-use")
Use fstat() on open file descriptor and use fcntl() to add
the desired O_NONBLOCK flag for pipes.
CoverityID: 442113 ("Resource leaks")
Ensure qctl allocated object is free'd in lfs_quota()
CoverityID: 442364: ("Null pointer dereferences")
In lfs_fid2path() ensure `path_or_fsname` is not null before
checking if it starts with a forward slash ('/')
CoverityID: 442116 ("Resource leaks")
In get_projid() close dir_fd as soon as it is no longer needed.
CoverityID: 442373 ("Memory - corruptions, overrun allocated")
Fix lov_forge_comp_v1() which also fails to clone FID
Test-Parameters: trivial
Test-Parameters: testlist=sanity env=ONLY="56ebb"
Signed-off-by: Shaun Tancheff <shaun.tancheff@hpe.com>
Change-Id: I2868d5ded322cd9cc890c463a494d296206d4be9
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56149
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Petros Koutoupis <petros.koutoupis@hpe.com>
Reviewed-by: Timothy Day <timday@amazon.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Etienne AUJAMES [Thu, 22 Aug 2024 15:46:19 +0000 (17:46 +0200)]
LU-18163 obdclass: fix sysfs_memparse()/string_to_size()
The patch reworks string_to_size() to avoid string copies and handle
default unit directly in __string_to_size().
Also, this would handle more gracefully fractional value with huge
unit (like 0.5EiB).
This implementation fixes the parsing of the following invalid
strings:
- "10.badbadMib"
- "10MiBbadbad"
- "10MBAD"
- "10.123badMib"
- "1024.KG"
The patch change the way to handle decimal fractional part: a maximum
of 9 digits are supported.
It fixes test_string_to_size_err() to actually return an error if a
test failed and then prevents the module to load.
It fixes obdclass_init() to avoid a crash if obd_init_checks()
failed.
Add regression tests in obd_init_checks.
Signed-off-by: Etienne AUJAMES <eaujames@ddn.com>
Change-Id: Ic20d11368fc7608637e8123d7c6c5a2ab2cf4a4b
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56135
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
Reviewed-by: Neil Brown <neilb@suse.de>
Reviewed-by: James Simmons <jsimmons@infradead.org>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Alex Zhuravlev [Tue, 4 Jun 2024 12:25:35 +0000 (15:25 +0300)]
LU-10026 ldiskfs: mballoc to preserve preallocation's start
.. used in dense preallocation. otherwise it's possible to lose
preallocated space when the corresponding cache bitmap is dropped
from the cache, then ldiskfs will be printing error messages
about block counter mismatch.
Fixes:
686dee707f ("LU-10026 osd-ldiskfs: use preallocation for dense writes")
Signed-off-by: Alex Zhuravlev <bzzz@whamcloud.com>
Change-Id: I93177510af959e849dba7a9c35d81bc27809a31b
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/55467
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Artem Blagodarenko <ablagodarenko@ddn.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Frederick Dilger [Fri, 14 Jun 2024 00:39:28 +0000 (20:39 -0400)]
LU-17432 libcfs: new CDEBUG_SLOW message type
Created new CDEBUG_SLOW message type that will skip the first
SECONDS of messages, then continue printing to console as normal.
Specifically CWARN_SLOW and CERROR_SLOW have been created which
mirror the functionality of their respective CWARN and CERROR
functions but with the additional SLOW behavior.
Signed-off-by: Frederick Dilger <fdilger@whamcloud.com>
Change-Id: I905fdff795488ff937faf4d04d5d3d6fec24950a
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/55439
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Arshad Hussain <arshad.hussain@aeoncomputing.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
Lai Siyao [Mon, 20 May 2024 21:20:34 +0000 (17:20 -0400)]
LU-17939 ofd: validate FID in MDT/OFD
OST object FID from other nodes can only be normal FID, IDIF, ECHO or
OST_MDT0, and MDT object FID can only be namespace visible FID or
local ROOT, return error if it's not true to avoid assertion later.
Signed-off-by: Lai Siyao <lai.siyao@whamcloud.com>
Change-Id: I0c518c67acad44e90159fff71ff4fa9b893e8f3d
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/55401
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Hongchao Zhang <hongchao@whamcloud.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
Mikhail Pershin [Tue, 28 May 2024 11:02:23 +0000 (14:02 +0300)]
LU-14510 dom: fiemap support for DoM files
Patch adds support for fiemap to DoM files.
Server part:
- modify MDS_GET_IMFO handler to return FIEMAP like
OST_GET_INFO does
- mdt_fiemap_get() to process fiemap request
Client part:
- rewrite lov_object_fiemap() to support DoM component
- rework fiemap_for_stripe() to work with both DoM and
RAID0 layouts
- use initialized layout entries to get subobject and
get rid of lov_find_subobj() used by fiemap only
- fix issue with wrong resume entry/stripe count
- mdc_object_fiemap() as implementation of .coo_fiemap
cl_object_operations to send and receive fiemap request
- treat LOV subdev errors as UNKNOWN extent
- rework FID2PATH layout description to be compatible with
other GET_INFO keys (no protocol changes)
- add sanity.sh test_130h for DoM fiemap with resuming
To indicate MDT device the extra bit is taken from stripe
number bits in favor of device number. So total absolute
stripe amount limit is 32768 in fiemap report
Signed-off-by: Mikhail Pershin <mpershin@whamcloud.com>
Change-Id: I9b6df04fd62d773aec2d916440ba08dfea06faa4
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/55221
Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
Reviewed-by: Lai Siyao <lai.siyao@whamcloud.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Jian Yu [Tue, 24 Sep 2024 07:29:19 +0000 (00:29 -0700)]
LU-15536 utils: add lfs somsync utility
This patch adds lfs somsync utility to synchronize
SOM xattr(s) for given FILE(s) or FID(s).
lfs somsync FILE ...
lfs somsync --by-fid MOUNT FID ...
Test-Parameters: trivial env=ONLY=807 testlist=sanity
Change-Id: Ie9ee39625d56ec026c89dcc0f27025904ca354e3
Signed-off-by: Jian Yu <yujian@whamcloud.com>
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/49498
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Alex Zhuravlev <bzzz@whamcloud.com>
Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Shaun Tancheff [Tue, 13 Aug 2024 14:10:34 +0000 (21:10 +0700)]
LU-18228 build: specify kmp-moddir
On debian 12 when packaging kernel integration test modules
the installed location needs to be well known.
Update debian/rules to specify kmp-moddir as well as adjust
the expected location of the kernel module.
HPE-bug-id: LUS-12444
Test-Parameters: trivial
Fixes:
bac5e458ee ("LU-17096 debian: add obd_test.ko, llog_test.ko to lustre-tests")
Signed-off-by: Shaun Tancheff <shaun.tancheff@hpe.com>
Change-Id: Ic431a78f372c6388907d1cc5db3ab39973cf31f0
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56386
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Reviewed-by: James Simmons <jsimmons@infradead.org>
Reviewed-by: Timothy Day <timday@amazon.com>
Reviewed-by: Petros Koutoupis <petros.koutoupis@hpe.com>
Andreas Dilger [Fri, 15 Oct 2021 02:31:48 +0000 (20:31 -0600)]
LU-13062 llite: return stripe_offset -1 in trusted.lov
If the trusted.lov xattr is copied by userspace to be restored later,
this results in PFL files always being restored onto OST0000 because
the kernel replaces the lmm_layout_gen field with 0 to avoid confusion
in userspace between the layout generation and the stripe offset (both
use the same field, one for input, one for output).
Instead of always returning 0 for the layout generation, return -1
for PFL layouts, so that restoring the xattr will direct the MDS to
select an appropriate OST.
For setxattr attempt patch prevent setting specific offsets if
PFL layout has an initialized entry what is considered as layout
copy attempt.
Signed-off-by: Andreas Dilger <adilger@whamcloud.com>
Change-Id: Ia1af2bfcfa41cf1593aab44fe2fa792c3d254035
Signed-off-by: Mikhail Pershin <mpershin@whamcloud.com>
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/45252
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Jian Yu <yujian@whamcloud.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Serguei Smirnov [Tue, 29 Oct 2024 20:36:14 +0000 (13:36 -0700)]
LU-18199 scripts: fix to always add local route
To avoid disruption of network connectivity during LNet start-up,
fix gateway selection logic in ksocklnd-config script so that
the local route is added along with the default route to the
custom routing table created by this script.
Another change in this patch makes sure that link-local IPv6
addresses are ignored, because they are ignored by LNet anyway
Fixes:
7f60b2b55 ("LU-17006 lnet: set up routes for going across subnets")
Test-Parameters: trivial testlist=sanity-lnet
Signed-off-by: Serguei Smirnov <ssmirnov@whamcloud.com>
Change-Id: I341a40268298709c772bad84ddf4a4fae645b48f
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56819
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Frank Sehr <fsehr@whamcloud.com>
Reviewed-by: Cyril Bordage <cbordage@whamcloud.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Mikhail Pershin [Sun, 8 Sep 2024 08:10:55 +0000 (11:10 +0300)]
LU-17906 pltrpc: don't use non-uptodate peer at connect
If peer is not yet discovered then LNET puts messages into
pending queue until discovery is done. That pins ptlrpc
request as well, thus a connect RPC to not alive peer is
stuck until peer discovery timed out despite RPC timeout.
Moreover that means no connect attempt to other peers are
made for that time:
nids_stats:
"192.168.252.112@tcp": { connects: 1, ... sec_ago: 31 }
"192.168.252.113@tcp": { connects: 0, ... sec_ago: never }
"192.168.252.115@tcp": { connects: 0, ... sec_ago: never }
After 30s it is still stuck with first NID and never tried
any other, despite connect RPC timeout is about 5-10s in
ptlrpc.
Patch prevents RPC stuck on non-uptodate peer just by
dropping such request in ptl_send_rpc(). That lets ptlrpc
to keep control over connection request expiration and new
connect attempts, so all peers are tried one by one until
some is ready.
Results with patch:
nids_stats:
"192.168.252.112@tcp": { connects: 4, ... sec_ago: 9 }
"192.168.252.113@tcp": { connects: 4, ... sec_ago: 4 }
"192.168.255.115@tcp": { connects: 3, ... sec_ago: 14 }
After the same 30s we had 11 connect attempts with all
failover NIDs tried
Patch modifies also LNetPeerDiscovered() to consider
a local peer as uptodate and return error code instead of
boolean.
Import uptodate state is also not boolen now but shows
discovery status.
Was-Change-Id: I51d8973aa8475ce1930f292c42aa22c70cfc13db
Test-Parameters: env=ONLY=153a,ONLY_REPEAT=10 testlist=conf-sanity
Test-Parameters: testlist=recovery-small,sanity-flr
Signed-off-by: Mikhail Pershin <mpershin@whamcloud.com>
Change-Id: I468faf6f4b0cf8ba0f4f810fe09a5f1165432d28
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56755
Reviewed-by: Chris Horn <chris.horn@hpe.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Josh Samuelson [Sat, 26 Oct 2024 20:06:37 +0000 (15:06 -0500)]
LU-18383 ksocklnd: Avoid TCP socket orphans in racy LNet
LU-18137
3367ef3eb2 introduced code to handle upstream kernel changes
with regards to releasing kernel sockets. Under heavy/racy LNet
hello transactions between servers and clients, client TCP sockets
can transition to a orphan state that is never reaped. This commit
moves LU-18137's fix closer to the kernel socket allocation to
avoid this racy connection setup condition.
This commit also provides a backport of the kernel's socket
accounting function sock_inuse_add() for compatibility when vendor
kernels lack the static function definition in their kernel headers.
The lnet_compat.h file has been introduced for a place to put such
compatibility code.
Fixes:
3367ef3eb2 ("LU-18137 ksocklnd: Fix TCP socket cleanup")
Signed-off-by: Josh Samuelson <josh@1up.unl.edu>
Change-Id: I11fc0b53e051871da26c079e9aeb50976b20bbd3
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56737
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Timothy Day <timday@amazon.com>
Reviewed-by: Chris Horn <chris.horn@hpe.com>
Reviewed-by: Mark Roper <ropermar@amazon.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Emoly Liu [Wed, 23 Oct 2024 00:53:00 +0000 (08:53 +0800)]
LU-17962 mgc: free nidlist correctly
Memory leak was found during interop testing because
nidlist was not freed correctly in function
mgc_apply_recover_logs().
Test-Parameters: testlist=conf-sanity env=ONLY=29 \
serverversion=2.15 clientdistro=el8.10 serverdistro=el8.10 \
mdscount=2 mdtcount=4 ostcount=8
Fixes:
e4d2d4ff74 ("LU-13306 mgc: handle large NID formats")
Signed-off-by: Emoly Liu <emoly@whamcloud.com>
Change-Id: I7a7a1b4b4f3b8c65c7608537e4dca1b9f1b68e77
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56709
Tested-by: Maloo <maloo@whamcloud.com>
Tested-by: jenkins <devops@whamcloud.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
Reviewed-by: James Simmons <jsimmons@infradead.org>
Chris Horn [Fri, 26 Jul 2024 20:36:51 +0000 (14:36 -0600)]
LU-14288 lnet: Introduce nidmasks
A nidmask is like a netmask, except it applies to IPv4 or IPv6 LNet
NIDs.
Nidmasks use the existing nidlist infrastructure so any caller of
cfs_parse_nidlist() can include a nidmask in the argument and match
NIDs against it using cfs_match_nid(), or convert it back to a string
with cfs_print_nidlist().
For example, "192.168.1.1@tcp/24" is equivalent to the nidrange
"192.168.1.[1-254]@tcp", and "2001::1@tcp/126" is equivalent to
"2001::@tcp 2001::1@tcp 2001::2@tcp 2001::3@tcp".
cfs_parse_nidrange() is modified to treat an IPv6 address as
equivalent to a netmask with prefix length of /128. Thus,
cfs_parse_nidlist() can now be used with lists of IPv6 addresses.
The user and kernel space implementations of cfs_parse_nidlist(),
et. al. have been modified to more closely match each other. Namely,
char * is used instead of the struct cfs_lstr and a length argument
is added to the kernel space cfs_parse_nidlist. Callers are adjusted
accordingly.
conf-sanity.sh/test_43a is modified to generate nidmasks that contain
the client's NID and verify that this is handled correctly when
nosquash_nids is set to the nidmask.
lnetctl debug nidlist command is added to facilitate testing of the
userspace code.
Test-Parameters: trivial testlist=conf-sanity env=ONLY=43a
Test-Parameters: testlist=conf-sanity env=ONLY=43a,FORCE_LARGE_NID=true,LOAD_MODULES_REMOTE=true
Signed-off-by: Chris Horn <chris.horn@hpe.com>
Change-Id: Id9d0bc6f4f8b977591f0b6f88bda46ae03cb58d5
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/55922
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Reviewed-by: Neil Brown <neilb@suse.de>
Reviewed-by: Serguei Smirnov <ssmirnov@whamcloud.com>
Alexey Lyashkov [Fri, 9 Aug 2024 08:33:42 +0000 (11:33 +0300)]
LU-18120 ldiskfs: improve a quota speedup patch
current patch might have a small window to make write duplicate,
lets avoid it.
HPe-bug-id: LUS-12432
Signed-off-by: Alexey Lyashkov <alexey.lyashkov@hpe.com>
Change-Id: I8de2e675d3bf5d5cbb41b25b04affe1d5e0d6411
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/55979
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Andrew Perepechko <andrew.perepechko@hpe.com>
Reviewed-by: Alex Zhuravlev <bzzz@whamcloud.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Bobi Jam [Wed, 28 Aug 2024 14:02:42 +0000 (22:02 +0800)]
LU-18176 idl: reserve OBD_CONNECT2_UPDATE_LAYOUT flags
Reserve some bits to be used in CSDC:
- OBD_CONNECT2_UPDATE_LAYOUT to negotiate new intent flag
- LAYOUT_INTENT_CHANGE to allow changing layout flags
- LAIF_INCOMPRESSIBLE to set LCME_FL_NOCOMPR flag in layout
Test-Parameters: trivial
Signed-off-by: Bobi Jam <bobijam@whamcloud.com>
Change-Id: Ifabeb6ce2669f9c88e1bba176b9ecc75e9b2b935
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56181
Reviewed-by: Artem Blagodarenko <ablagodarenko@ddn.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Yang Sheng [Tue, 10 Sep 2024 16:14:21 +0000 (00:14 +0800)]
LU-18214 ldlm: change flock deadlock detection
The flock deadlock detection code thought request lock
same as blocking lock is a bug. In fact, this is a case
of cycling chain. So we should treat it as a deadlock
case. Also clean up the reprocess code.
Signed-off-by: Yang Sheng <ys@whamcloud.com>
Change-Id: Icf0df4ac281c2cdb6cc57cb79db137d39ecef9e6
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56319
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Mikhail Pershin <mpershin@whamcloud.com>
Reviewed-by: Alex Zhuravlev <bzzz@whamcloud.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Oleg Drokin [Tue, 12 Nov 2024 05:16:33 +0000 (00:16 -0500)]
Post-branch-off of 2.16 master is now 2.17
Change-Id: I15deb804e67969e7fd97fb3620f78ec835cfd51a
Signed-off-by: Oleg Drokin <green@whamcloud.com>
Oleg Drokin [Fri, 8 Nov 2024 20:45:12 +0000 (15:45 -0500)]
New release 2.16.0
Change-Id: I8059e5e445f200fb77600213c0d0b4c2046ee2ae
Signed-off-by: Oleg Drokin <green@whamcloud.com>
Andreas Dilger [Thu, 7 Nov 2024 04:49:52 +0000 (21:49 -0700)]
LU-17251 tests: force test_rr_alloc new seq
Force parallel-scale test_rr_alloc to create a new OST sequence
if the number of precreated objects is low. This should ensure
that the number of available objects is enough to complete the
test, and avoids OST object creation being blocked in the middle
of the test while the SEQ number rolls over.
Test-Parameters: trivial
Test-Parameters: testgroup=review-dne-part-9 env=RACER_EXCEPT="1 2"
Test-Parameters: testgroup=review-dne-part-9 env=RACER_EXCEPT="1 2"
Test-Parameters: testgroup=review-dne-part-9 env=RACER_EXCEPT="1 2"
Test-Parameters: testgroup=review-dne-part-9 env=RACER_EXCEPT="1 2"
Test-Parameters: testgroup=review-dne-part-9 env=RACER_EXCEPT="1 2"
Test-Parameters: testgroup=review-dne-part-9 env=RACER_EXCEPT="1 2"
Test-Parameters: testgroup=review-dne-part-9 env=RACER_EXCEPT="1 2"
Test-Parameters: testgroup=review-dne-part-9 env=RACER_EXCEPT="1 2"
Test-Parameters: testgroup=review-dne-part-9 env=RACER_EXCEPT="1 2"
Test-Parameters: testgroup=review-dne-part-9 env=RACER_EXCEPT="1 2"
Test-Parameters: testlist=parallel-scale env=ONLY=rr_alloc,ONLY_REPEAT=50
Signed-off-by: Andreas Dilger <adilger@whamcloud.com>
Change-Id: I636a488e575d27ac235749911f171d5e1e3ebbe5
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56913
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Li Dongyang <dongyangli@ddn.com>
Reviewed-by: Jian Yu <yujian@whamcloud.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Andreas Dilger [Mon, 4 Nov 2024 20:01:56 +0000 (12:01 -0800)]
LU-17251 tests: try to fix test_rr_alloc again
Try to fix parallel-scale test_rr_alloc again:
- ensure that the test directory is striped over all MDTs to
maximize the number of precreated objects available
- ensure the FID SEQ has OIDs to not run out during this test,
which caused some OSPs to run out of objects during creation
Add debugging to understand issue more if it continues to fail:
- print lfs df, lfs df -i at start and error to show imbalance
- delete files only in cleanup_rr_alloc() so distribution can be shown
Clean up test script to ensue SEQ is large enough to run test.
Test-Parameters: trivial
Test-Parameters: testgroup=review-dne-part-9 env=RACER_EXCEPT="1 2"
Test-Parameters: testgroup=review-dne-part-9 env=RACER_EXCEPT="1 2"
Test-Parameters: testgroup=review-dne-part-9 env=RACER_EXCEPT="1 2"
Test-Parameters: testgroup=review-dne-part-9 env=RACER_EXCEPT="1 2"
Test-Parameters: testgroup=review-dne-part-9 env=RACER_EXCEPT="1 2"
Test-Parameters: testgroup=review-dne-part-9 env=RACER_EXCEPT="1 2"
Test-Parameters: testgroup=review-dne-part-9 env=RACER_EXCEPT="1 2"
Test-Parameters: testgroup=review-dne-part-9 env=RACER_EXCEPT="1 2"
Test-Parameters: testgroup=review-dne-part-9 env=RACER_EXCEPT="1 2"
Test-Parameters: testgroup=review-dne-part-9 env=RACER_EXCEPT="1 2"
Test-Parameters: testlist=parallel-scale env=ONLY=rr_alloc,ONLY_REPEAT=50
Signed-off-by: Andreas Dilger <adilger@whamcloud.com>
Change-Id: I636a488e575d27ac235749911f171d5e1e33e310
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56853
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Jian Yu <yujian@whamcloud.com>
Reviewed-by: Li Dongyang <dongyangli@ddn.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Andreas Dilger [Tue, 5 Nov 2024 00:42:38 +0000 (16:42 -0800)]
LU-17712 tests: allow multiop to return 0 in recovery-small/157
This patch allows multiop to return 0 in recovery-small/157,
which is conserdered as success because the mmaped I/O
did not hang after the client was evicted.
The patch also skips recovery-small/157 for older MDS since
the test depends on changes made to the MDS code on the server.
Test-Parameters: trivial testlist=recovery-small serverversion=2.15.5
Fixes:
71f8e5d650 ("LU-14708 ptlrpc: skip unnecessary client eviction")
Change-Id: I19d0362ffbccb44c81ec2e09d9b4eccba40b9dcf
Signed-off-by: Andreas Dilger <adilger@whamcloud.com>
Signed-off-by: Jian Yu <yujian@whamcloud.com>
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56884
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Emoly Liu <emoly@whamcloud.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Hongchao Zhang [Sat, 5 Oct 2024 16:12:08 +0000 (00:12 +0800)]
LU-18394 test: adjust step for different OS Arch
in test_49 of sanity-quota, the Bash could fail if the data to be
processsed by "eval" is big under PPC64le, decrease the data size
to be suitable for different Arches.
Test-Parameters: trivial testlist=sanity-quota env=SLOW=yes,ONLY=49,ONLY_REPEAT=10 clientdistro=el8.9 clientarch=ppc64le serverdistro=el8.8
Signed-off-by: Hongchao Zhang <hongchao@whamcloud.com>
Change-Id: I6fef841211662c3518caf97835598dc26beaea1f
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56820
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Jian Yu <yujian@whamcloud.com>
Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Andreas Dilger [Thu, 10 Oct 2024 02:01:25 +0000 (20:01 -0600)]
LU-18288 tests: lru_resize_disable sets lru_max_age
sanity test_120* has started failing much more regularly since the
lru_max_age=600s was made the default. This is caused by DLM locks
being aged out of the LRU during the test and confuses the result.
Set lru_max_age to the old (65 min) limit in lru_resize_disable()
for tests that don't want locks to be cancelled during the subtest.
Register a stack_trap in lru_resize_disable() to reset lru_max_age
to the old value, so that the caller does not need to remember this.
Resetting lru_size to the "old" value cannot be done directly, since
it returns the number of locks in the LRU, and not the LRU size limit.
Instead, register lru_resize_enable() with stack_trap() to reset it.
Add debugging to test cases that were failing with earlier versions
of this patch, to help understand similar failures in the future.
Script code style fixes in modified subtests.
Fixes:
357cae970c ("LU-17428 ldlm: reduce default lru_max_age")
Signed-off-by: Andreas Dilger <adilger@whamcloud.com>
Change-Id: I82ee177eae14f3030a9e92e3aca86e4c47401ff5
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56642
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Timothy Day <timday@amazon.com>
Reviewed-by: Zhenyu Xu <bobijam@hotmail.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Jian Yu [Fri, 1 Nov 2024 16:57:38 +0000 (09:57 -0700)]
LU-18402 tests: skip recovery-small/{153,154a} for older MDS
Skip recovery-small/{153,154a} for older MDS since
the test depends on changes made to the MDS code on the server.
Test-Parameters: trivial testlist=recovery-small serverversion=2.15.5
Fixes:
654d5f3fa4 ("LU-16478 target: disconnected export")
Fixes:
e818052444 ("LU-17365 lod: handle llog errors gracefuly")
Signed-off-by: Jian Yu <yujian@whamcloud.com>
Change-Id: I3f26fae9d8d3d338f0e0860ea10f5fb762d70640
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56829
Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
Reviewed-by: Emoly Liu <emoly@whamcloud.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Serguei Smirnov [Thu, 31 Oct 2024 00:04:51 +0000 (17:04 -0700)]
LU-18320 tests: add skip option to sanity-lnet test_226
Update sanity-lnet test_226 to be skipped if peer doesn't
have the refcount fix from LU-17440.
Test-Parameters: trivial testlist=sanity-lnet
Fixes:
2b210f3905 ("LU-17440 lnet: prevent errorneous decref for asym route")
Signed-off-by: Serguei Smirnov <ssmirnov@whamcloud.com>
Change-Id: I274668048f3df27da1b226e9fb1966dbbd877713
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56839
Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
Reviewed-by: Frank Sehr <fsehr@whamcloud.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Andreas Dilger [Wed, 30 Oct 2024 22:10:56 +0000 (16:10 -0600)]
LU-17527 tests: fix sanity/255a decimal comparison
Don't use fractional percentages in the sanity test_255a performance
comparison, since bash (( ... )) cannot compare numbers with decimal
points properly. Instead, just compute the percentage speedup with
whole numbers, since the test discards anything less than 20% speedup
and a fraction of a percent will not make much difference here.
Test-Parameters: trivial
Fixes:
bdd470ff97 ("LU-9069 tests: improve output of sanity test_255a")
Fixes:
395f3e1a55 ("LU-15316 tests: use integers in sanity test_255a")
Signed-off-by: Andreas Dilger <adilger@whamcloud.com>
Change-Id: Id3b37e07168ee2590e52d01f66336027254ced55
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56836
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Arshad Hussain <arshad.hussain@aeoncomputing.com>
Reviewed-by: Jian Yu <yujian@whamcloud.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Charlie Olmstead [Tue, 1 Oct 2024 17:02:05 +0000 (11:02 -0600)]
LU-18286 tests: auster node.yml labels rocky9 as RHEL
release() assumes a node with /etc/redhat-release is RHEL.
This patch removes reading this file in favor of os-release.
Reading of centos-release (if present) still required as
os-release for centos distros doesn't include minor version.
Test-Parameters: trivial
Signed-off-by: Charlie Olmstead <charlie@whamcloud.com>
Change-Id: I888f8eeacaf843120b2beb134292047b3907a9a6
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56580
Tested-by: Maloo <maloo@whamcloud.com>
Tested-by: jenkins <devops@whamcloud.com>
Reviewed-by: Alex Deiter <adeiter@ddn.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Andreas Dilger [Fri, 4 Oct 2024 07:15:03 +0000 (01:15 -0600)]
LU-18310 tests: add debugging to test_metabench
Both parallel-scale and parallel-scale-nfs are intermittently
failing test_metabench with "No space left on device" (ENOSPC)
or "Disk quota exceeded" (EDQUOT), even though this test is
creating only about 10-20k files.
Add some debugging to see where all of the space has gone, and
what quota limits are being set. It may be that some earlier
test (e.g. compilebench) is leaving too much junk behind.
The failure rate is very low (only 2/637 runs in the past 4 weeks),
so it likely needs to be landed to catch a failure.
Test-Parameters: trivial testlist=parallel-scale
Test-Parameters: testlist=parallel-scale-nfsv4
Test-Parameters: testgroup=full-part-1
Signed-off-by: Andreas Dilger <adilger@whamcloud.com>
Change-Id: Ie35ae677032ccc8113cbad5dc5a7b0504149717f
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56582
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Alex Deiter <adeiter@ddn.com>
Reviewed-by: Elena <elena.gryaznova@hpe.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Qian Yingjin [Tue, 29 Oct 2024 03:25:33 +0000 (11:25 +0800)]
LU-18298 tests: skip sanity-pcc/test_1{c,d} on SLES15 SP5
Skip the sanity-pcc/test_1{c, d} failure on SLES15 SP3 - SP5.
However, they passed on SLES15 SP6.
Test-Parameters: trivial
Signed-off-by: Qian Yingjin <qian@ddn.com>
Change-Id: I08fd0a192307f0072cc82033958dd8239ea507d5
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56809
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Tested-by: Jian Yu <yujian@whamcloud.com>
Reviewed-by: Jian Yu <yujian@whamcloud.com>
Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Emoly Liu [Fri, 25 Oct 2024 07:05:33 +0000 (15:05 +0800)]
LU-18393 tests: $num_files should be multiple of $num_entries
According to performance-sanity.sh test_4 failure
"md_validate_tests, items must be a multiple of items per directory",
set $num_files to be a multiple of $num_entries.
Test-Parameters: trivial testlist=performance-sanity
Signed-off-by: Emoly Liu <emoly@whamcloud.com>
Change-Id: I8c635649ef016389d1bd22f8318a55f8d0f77962
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56785
Tested-by: Maloo <maloo@whamcloud.com>
Tested-by: jenkins <devops@whamcloud.com>
Reviewed-by: Alex Deiter <adeiter@ddn.com>
Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Alex Zhuravlev [Thu, 17 Oct 2024 18:22:11 +0000 (21:22 +0300)]
LU-18379 tests: wait for stats to expire in conf-sanity/127
fs stats are not updated immediately on the client,
thus we need to wait.
Test-Parameters: env=ONLY=127,ONLY_REPEAT=20 testlist=conf-sanity
Test-Parameters: env=ONLY=127,ONLY_REPEAT=20 testlist=conf-sanity
Signed-off-by: Alex Zhuravlev <bzzz@whamcloud.com>
Change-Id: I28cf407f9fe4df1f46af8cd88f50670bb8f0d93f
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56726
Tested-by: Maloo <maloo@whamcloud.com>
Tested-by: jenkins <devops@whamcloud.com>
Reviewed-by: Jian Yu <yujian@whamcloud.com>
Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Jian Yu [Sat, 26 Oct 2024 03:56:45 +0000 (20:56 -0700)]
LU-18402 tests: skip recovery-small/155 for older MDS
Skip recovery-small test 155 for MDS < 2.15.58.110 since
the test depends on changes made to the MDS code on the server.
Test-Parameters: trivial testlist=recovery-small env=ONLY="155 157" \
serverversion=2.15.5
Fixes:
71f8e5d650 ("LU-14708 ptlrpc: skip unnecessary client eviction")
Change-Id: I44137ebbfb1ec0f9a6a1cf1b42cd211caa146009
Signed-off-by: Jian Yu <yujian@whamcloud.com>
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56795
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
Reviewed-by: Mikhail Pershin <mpershin@whamcloud.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Oleg Drokin [Sat, 26 Oct 2024 23:58:55 +0000 (19:58 -0400)]
New RC 2.16.0-RC5
Change-Id: I677f9f232df161eb37fc917469a6c23f48716b55
Signed-off-by: Oleg Drokin <green@whamcloud.com>
Shaun Tancheff [Thu, 24 Oct 2024 06:58:41 +0000 (13:58 +0700)]
LU-17525 llite: soft fail unaligned dio
Skip unaligned DIO and pass io as buffered I/O rather than
hard fail unaligned DIO.
Test-Parameters: testlist=sanity serverversion=2.14 env=SANITY_EXCEPT="65n 211 413"
Fixes:
ff018bb77a ("LU-18284 llite: disallow udio exceptions")
Signed-off-by: Shaun Tancheff <shaun.tancheff@hpe.com>
Change-Id: I638f62ec96abc3032da5fbcf895cd835022fd759
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56775
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
Reviewed-by: Patrick Farrell <patrick.farrell@oracle.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Andreas Dilger [Fri, 25 Oct 2024 07:54:01 +0000 (01:54 -0600)]
LU-18389 tests: skip added sanity/65n checks
Skip checks added at the end of sanity test_65n for MDS versions
that do not have the new layout inheritance behavior.
Test-Parameters: trivial env=ONLY=65n testlist=sanity serverversion=2.14
Fixes:
6e59408f1a ("LU-12130 lod: make pool inheritance policy more consistent")
Signed-off-by: Andreas Dilger <adilger@whamcloud.com>
Change-Id: Ic7c36801ec6e906d631bc4fc234f1f2b77e9f7dc
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56786
Reviewed-by: Jian Yu <yujian@whamcloud.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Tested-by: jenkins <devops@whamcloud.com>
Vladimir Saveliev [Wed, 16 Oct 2024 14:45:37 +0000 (17:45 +0300)]
LU-17893 tests: wait destroys before replay-dual:test_28
replay-dual.sh:test_28() should take care that it drops only own
blocking ast. If test_26() ran before, there may be pending destroys
when test_28() runs. Dropping of blocking asts for destroys makes
replay-dual.sh to get accompanied with:
watchdog stack traces:
[169376.453554] Lustre: ll_ost00_057: service thread pid 236757 was
inactive for 40.816 seconds. Watchdog stack traces are limited to 3
per 300 seconds, skipping this one.
[169376.461659] [<0>] ldlm_completion_ast+0x99b/0xc00 [ptlrpc]
[169376.461782] [<0>] ldlm_cli_enqueue_local+0x302/0x890 [ptlrpc]
[169376.461888] [<0>] ofd_destroy_by_fid+0x29c/0x570 [ofd]
[169376.461906] [<0>] ofd_destroy_hdl+0x22c/0x960 [ofd]
lock timeouts:
[169638.155933] LustreError:
236757:0:(ldlm_request.c:104:ldlm_expired_completion_wait()) ###
lock timed out (enqueued at
1729087746, 303s ago); not entering
recovery in server code, just going back to sleep ns..
and system overload indications:
[169852.021044] Lustre: ll_ost00_052: service thread pid 236555
completed after 516.964s. This likely indicates the system was
overloaded (too many service threads, or not enough hardware
resources).
Wait for completion of destroys before starting test_28().
Test-Parameters: trivial testlist=replay-dual
Signed-off-by: Vladimir Saveliev <vladimir.saveliev@hpe.com>
Change-Id: I837579a428d8c2383fe884961d356ff417fc3f2e
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56712
Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
Reviewed-by: Jian Yu <yujian@whamcloud.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Tested-by: jenkins <devops@whamcloud.com>
Andreas Dilger [Tue, 1 Oct 2024 21:06:06 +0000 (15:06 -0600)]
LU-14330 tests: wait for orphan thread to exit
It may take a few seconds for the orphan cleanup thread to finish.
Wait for the thread to exit rather than failing the test.
Test-Parameters: trivial
Fixes:
a1e6e75a82 ("LU-12846 tests: verify orphan upgrade compatibilty")
Signed-off-by: Andreas Dilger <adilger@whamcloud.com>
Change-Id: I604be1b9f8f460d9183ba1aaddd3b77e153ebbe5
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56559
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Artem Blagodarenko <ablagodarenko@ddn.com>
Reviewed-by: Alex Deiter <adeiter@ddn.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Lai Siyao [Sat, 8 Jul 2023 22:32:29 +0000 (18:32 -0400)]
LU-15553 test: mkdir_on_mdt0 in replay-dual
Several subtests in replay-dual require test dir created on MDT0,
replace mkdir with mkdir_on_mdt0. These subtests are found by script:
grep -C 10 -n "do_facet.*SINGLEMDS" lustre/tests/*.sh | grep -w mkdir
Fixes:
b9c4dc3c33 ("LU-14792 llite: enable filesystem-wide default LMV")
Test-Parameters: trivial mdscount=2 mdtcount=4 testlist=replay-dual,replay-dual,replay-dual
Signed-off-by: Lai Siyao <lai.siyao@whamcloud.com>
Change-Id: Ib28cf35575546c61bb7fa1b2c8a87ac31bd1ad4e
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/51665
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
Reviewed-by: Jian Yu <yujian@whamcloud.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Lai Siyao [Mon, 23 Sep 2024 01:10:20 +0000 (21:10 -0400)]
LU-15553 test: mkdir_on_mdt0 in conf-sanity.sh
Change mkdir to mkdir_on_mdt0 in several conf-sanity.sh sub tests.
Fixes:
b9c4dc3c33 ("LU-14792 llite: enable filesystem-wide default LMV")
Test-Parameters: trivial testlist=conf-sanity mdtcount=4
Signed-off-by: Lai Siyao <lai.siyao@whamcloud.com>
Change-Id: I5ace9df10e725802ba502ca20c60afc708b857cc
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56539
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
Reviewed-by: Jian Yu <yujian@whamcloud.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Lai Siyao [Mon, 23 Sep 2024 01:17:40 +0000 (21:17 -0400)]
LU-15553 test: mkdir_on_mdt0 in replay-vbr.sh
Change mkdir to mkdir_on_mdt0 in several replay-vbr.sh sub tests.
Fixes:
b9c4dc3c33 ("LU-14792 llite: enable filesystem-wide default LMV")
Test-Parameters: trivial testlist=replay-vbr mdtcount=4
Signed-off-by: Lai Siyao <lai.siyao@whamcloud.com>
Change-Id: I7457c155bbadb86adf8272113a4e4202b98c20a5
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56540
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
Reviewed-by: Jian Yu <yujian@whamcloud.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Arshad Hussain [Wed, 23 Oct 2024 11:08:52 +0000 (07:08 -0400)]
LU-16870 tests: Make 413a/b work on server 2.14 with client master
This patch makes 413a/b interop work on server 2.14
with client master.
First, this patch removes general redirect of cmd execution
(fallocate/dd) to dev/null from generate_uneven_mdts(). For
'dd' it might get little verbose however on failure it will
dump output to stdout giving more info.
Second, it adds check under check_fallocate_supported() to check
is fallocate is issued on MDS. If yes it check for version when
this feature was added. If not it falls back to 'dd' command
Third, under unload_modules() it check for version where
unload_modules_local() was added. If not it falls back to
non-unload_modules_local() version to unload module.
Test-Parameters: trivial testlist=sanity serverversion=2.14 env=ONLY=413a,413b,ONLY_REPEAT=10
Signed-off-by: Arshad Hussain <arshad.hussain@aeoncomputing.com>
Change-Id: I8a8843dd32f7e88d6d0938b67ce24353c9f9cb65
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56768
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
Reviewed-by: Alex Deiter <adeiter@ddn.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Hongchao Zhang [Sun, 3 Mar 2024 02:11:36 +0000 (10:11 +0800)]
LU-18024 quota: fix the order of freeing qmt_lvbo_free_wq
In qmt_device_fini, put the freeing of the qmt_lvbo_free_wq after
all possible usage of it having finished cleanup.
Signed-off-by: Hongchao Zhang <hongchao@whamcloud.com>
Change-Id: Ia26f1bc490adff5ae7dc850e2c89baf5874f01c5
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56778
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Sergey Cheremencev <scherementsev@ddn.com>
Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Bobi Jam [Tue, 22 Oct 2024 02:15:12 +0000 (10:15 +0800)]
LU-17670 tests: sanity-flr/33b allow bigger time difference
Allow a bigger time difference between read that switching among
mirrors.
Test-Parameters: trivial testlist=sanity-flr env=ONLY=33b
Signed-off-by: Bobi Jam <bobijam@whamcloud.com>
Change-Id: I14482e290fe5a96d1d10b2d47c8b4cfcadec96ec
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56752
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
Reviewed-by: Colin Faber <cfaber@ddn.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Lai Siyao [Sun, 29 Sep 2024 16:31:07 +0000 (12:31 -0400)]
LU-18343 lod: old client mkdir on wrong MDT
Old clients may not cache default LMV, and mkdir request may be sent
to wrong MDT, but MDT is too strict. Return -EREMOTE instead of
-EPROTO to let old clients retry.
Fixes:
6dbb4c6c8 ("LU-12998 lod: statfs upon nocreate check")
Test-Parameters: mdtcount=4 clientversion=2.15 testlist=sanity env=ONLY=160a
Signed-off-by: Lai Siyao <lai.siyao@whamcloud.com>
Change-Id: I2e64e2b3ccd3117de163f58d697191fab4d7be1c
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56731
Reviewed-by: Emoly Liu <emoly@whamcloud.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Tested-by: jenkins <devops@whamcloud.com>
Feng Lei [Wed, 16 Oct 2024 08:21:22 +0000 (16:21 +0800)]
LU-18329 tests: check version for parallel-scale-nfs/test_1
Check the lustre client version of nfs server (MDS1 by default) in
parallel-scale-nfs/test_1 for interop test.
Fixes:
69dcd1b940 ("LU-18030 tests: Add a test to ensure permissions copy on nfs")
Signed-off-by: Feng Lei <flei@whamcloud.com>
Test-Parameters: trivial
Test-Parameters: testlist=parallel-scale-nfsv3 env=ONLY=1
Test-Parameters: serverversion=2.15.5 testlist=parallel-scale-nfsv3 env=ONLY=1
Change-Id: I76ecb3bc28f37ba7d0c24d18eead621d6b066800
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56674
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
Reviewed-by: Jian Yu <yujian@whamcloud.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Oleg Drokin [Tue, 22 Oct 2024 00:27:19 +0000 (20:27 -0400)]
New tag 2.16.0-RC4
Change-Id: I9f87bbf152d51f50d1df6c9e75c0d1827a4e40a8
Signed-off-by: Oleg Drokin <green@whamcloud.com>
Jian Yu [Sat, 19 Oct 2024 00:11:09 +0000 (17:11 -0700)]
LU-17906 pltrpc: Revert "LU-17906 pltrpc: don't use non-uptodate peer at connect"
This reverts commit
6fe522d3d4f92aa2a48a573419f4590b10ef13d3.
The commit caused many regression failures: LU-18365, LU-18367,
LU-18368, LU-18366, etc.
Change-Id: Iaf7a0ec7606dab7a1b4d5b9f4fd1a24eb2c8d94d
Signed-off-by: Jian Yu <yujian@whamcloud.com>
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56736
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Oleg Drokin <green@whamcloud.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Oleg Drokin [Fri, 18 Oct 2024 18:53:04 +0000 (14:53 -0400)]
New RC 2.16.0-RC3
Change-Id: I89424c8fe80025746cc5baada8782ae2c77a2530
Signed-off-by: Oleg Drokin <green@whamcloud.com>
Hongchao Zhang [Sun, 29 Sep 2024 14:16:42 +0000 (22:16 +0800)]
LU-18024 quota: relate qmt_lvbo_free_wq and QMT
The global variable "qmt_lvbo_free_wq" could be affected if there
are more than one Lustre mount (for instance, using snapshot),
this patch moves it into "struct qmt_device" and it will only be
used by this QMT(MDT0000) and avoid conflict.
Fixes:
2cc18ece1e ("LU-14535 quota: free lvbo in a wq")
Signed-off-by: Hongchao Zhang <hongchao@whamcloud.com>
Change-Id: I298382fa0eed6885e55991e26e4b36d435e99e26
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56661
Tested-by: Maloo <maloo@whamcloud.com>
Tested-by: jenkins <devops@whamcloud.com>
Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Reviewed-by: Sergey Cheremencev <scherementsev@ddn.com>
Jian Yu [Tue, 15 Oct 2024 19:47:24 +0000 (12:47 -0700)]
LU-18324 tests: skip sanity/205k for older MDS
Skip sanity test 205k for MDS < 2.15.64.85 since the test
depends on changes made to the MDS code on the server.
Test-Parameters: trivial testlist=sanity env=ONLY=205k \
serverversion=2.15.5
Change-Id: I23709c2414b16a3d3a1ec742e7213597183b8f2a
Fixes:
ecdcaa3986 ("LU-17512 utils: new ? operator for jobid_name")
Signed-off-by: Jian Yu <yujian@whamcloud.com>
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56703
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
Reviewed-by: Colin Faber <cfaber@ddn.com>
Reviewed-by: Sarah Liu <sarah@whamcloud.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Jian Yu [Fri, 11 Oct 2024 06:01:40 +0000 (23:01 -0700)]
LU-18322 tests: skip recovery-small/154b for older MDS
Skip recovery-small test 154b for MDS < 2.15.60.2 since
the test depends on changes made to the MDS code on the server.
Test-Parameters: trivial testlist=recovery-small env=ONLY=154b \
serverversion=2.15.5
Change-Id: I540b943abae2c39cbb62d5403e37cd7768cb96b2
Fixes:
e818052444 ("LU-17365 lod: handle llog errors gracefuly")
Signed-off-by: Jian Yu <yujian@whamcloud.com>
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56658
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
Reviewed-by: Emoly Liu <emoly@whamcloud.com>
Reviewed-by: Mikhail Pershin <mpershin@whamcloud.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Qian Yingjin [Mon, 14 Oct 2024 02:43:31 +0000 (10:43 +0800)]
LU-18330 tests: add interop PCCRO check for sanity-pcc/test_35
Add interop PCCRO connection flag check for sanity-pcc/
test_{35,36}.
Fix the interop sanity-pcc tests with b2_15.
Test-Parameters: trivial serverversion=2.15.5 testlist=sanity-pcc
Test-Parameters: trivial testlist=sanity-pcc env=ONLY=36b,ONLY_REPEAT=100
Signed-off-by: Qian Yingjin <qian@ddn.com>
Change-Id: I9174752bf123abf608a2d3e1f72650cf08e7146e
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56675
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
Reviewed-by: James Simmons <jsimmons@infradead.org>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Bobi Jam [Mon, 14 Oct 2024 04:05:13 +0000 (12:05 +0800)]
LU-18319 tests: skip sanity-flr/test_50a for old MDS
Skip sanity-flr test_50a for old server because of changes that
were made on master from LU-13805 to reject unaligned DIO.
Test-Parameters: trivial
Fixes:
88d324be08 ("LU-13805 osd: Implement unaligned DIO connect flag")
Signed-off-by: Bobi Jam <bobijam@whamcloud.com>
Change-Id: I1bbd7528de458b4cca48c466203fc5cdd0ca4e8e
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56676
Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
Reviewed-by: Colin Faber <cfaber@ddn.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Yang Sheng [Wed, 9 Oct 2024 15:13:27 +0000 (23:13 +0800)]
LU-18325 tests: sanity 842: load_module ldlm_extent failed
Skip the MDS without ldlm_extent module.
Test-Parameters: trivial testlist=sanity
Signed-off-by: Yang Sheng <ys@whamcloud.com>
Change-Id: I9d1ec659088b7314ee3ae2fb09fc49ee3ef5a922
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56630
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
Reviewed-by: Jian Yu <yujian@whamcloud.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Ronnie Sahlberg [Thu, 10 Oct 2024 03:57:39 +0000 (23:57 -0400)]
LU-17047 tests: sanity/77c ensure ost debug_path is valid
test_77c has been failing intermittently for a while due
to either debug_param not being set properly before this test is
invoked or due to silent failure of getting the param from the
remote node.
Try to avoid these spurios test flakiness by forcing the debug_path
parameter to the expected setting during test startup.
A test shouldn't depend on no residuals being present from unrelated
prior tests anyway.
If need be we can add a dedicated test that only verifies that we
can set debug_path to a specific value and then read the value back.
Test-Parameters: trivial testgroup=rolling-upgrade-client1
Test-Parameters: testgroup=rolling-downgrade-mds
Test-Parameters: testgroup=rolling-downgrade-client2
Test-Parameters: testgroup=rolling-upgrade-client2
Signed-off-by: Ronnie Sahlberg <rsahlberg@whamcloud.com>
Change-Id: I4aae3e422d8f492a7877fbe20a6c553afc35e998
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56643
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Jian Yu <yujian@whamcloud.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
Andreas Dilger [Thu, 10 Oct 2024 00:22:37 +0000 (18:22 -0600)]
LU-18335 tests: skip conf-sanity/136 in interop
conf-sanity test_136 was added to test the fix for a bug
that crashes the MDS, so skip it during interop testing.
Test-Parameters: trivial testlist=conf-sanity env=ONLY=136
Test-Parameters: testlist=conf-sanity env=ONLY=136 serverversion=2.15
Fixes:
55c143a66d ("LU-15784 obdecho: don't panic with run on second mdt")
Signed-off-by: Andreas Dilger <adilger@whamcloud.com>
Change-Id: Iec0f96387bf69218ee74b88cce5ab21e07004292
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56638
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Jian Yu <yujian@whamcloud.com>
Reviewed-by: Minh Diep <mdiep@whamcloud.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Jian Yu [Wed, 9 Oct 2024 05:47:22 +0000 (22:47 -0700)]
LU-18326 tests: skip sanity-hsm/410 for older MDS
Skip sanity-hsm test 410 for MDS < 2.15.90.10 since the test depends on
changes made to the MDS code on the server.
Test-Parameters: trivial testlist=sanity-hsm env=ONLY=410 serverversion=2.15.5
Change-Id: Ib4471b6b9e3370fe021cef1b4a37f3338bc118c1
Fixes:
80a961261a ("LU-17863 hsm: add ioctl to set data version in HSM xattr")
Signed-off-by: Jian Yu <yujian@whamcloud.com>
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56623
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
Reviewed-by: Li Xi <lixi@ddn.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Jian Yu [Tue, 8 Oct 2024 23:56:05 +0000 (16:56 -0700)]
LU-18323 tests: skip sanity/27T for OSTs < 2.15.65.18
Skip sanity test_27T for OSTs < 2.15.65.18 since the test depends on
changes made to the OST code on the server.
Test-Parameters: trivial testlist=sanity env=ONLY=27T serverversion=2.15.5
Change-Id: I96b134f3ade1d417adf751a69cdcfc71c15b4d75
Fixes:
8a865c65c6 ("LU-17567 tests: improve sanity.sh:test_27T")
Fixes:
ccaff8d3c0 ("LU-17567 tests: skip sanity/27T for older OSTs")
Signed-off-by: Jian Yu <yujian@whamcloud.com>
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56615
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
Reviewed-by: Sarah Liu <sarah@whamcloud.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Jian Yu [Sat, 12 Oct 2024 01:39:43 +0000 (18:39 -0700)]
LU-18321 tests: skip ost-pools/31 for older MDS
Skip the loop spilling testing in ost-pools test 31 for
MDS < 2.15.50.150 since the test depends on changes made
to the MDS code on the server.
Test-Parameters: trivial testlist=ost-pools env=ONLY=31 \
serverversion=2.15.5
Change-Id: I4ad2ded8073eca55363fe60cc49c580ca7d983ae
Fixes:
c9c842d678 ("LU-15043 lod: check for spilling loops")
Signed-off-by: Jian Yu <yujian@whamcloud.com>
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56657
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
Reviewed-by: Sarah Liu <sarah@whamcloud.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Andreas Dilger [Wed, 9 Oct 2024 23:33:16 +0000 (17:33 -0600)]
LU-18226 tests: allow small margin for unevicted pages
Allow a small margin for unevictable pages in case of memory pressure.
Test-Parameters: trivial
Fixes:
2a3ec2fea3 ("LU-17463 osc: add support for unevictable mlock()ed pages")
Signed-off-by: Andreas Dilger <adilger@whamcloud.com>
Change-Id: I3f1cbacfdf53f217d9ebcbeb1b34693d27f70901
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56637
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Jian Yu <yujian@whamcloud.com>
Reviewed-by: Colin Faber <cfaber@ddn.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Hongchao Zhang [Sun, 29 Sep 2024 00:32:39 +0000 (08:32 +0800)]
LU-18293 test: check to sync quota setting
If some quota setting version is droped between QMT and QSD,
the quota setting could not be transfered to QSD and cause
the corresponding quota ID is not enforced properly.
Test-Parameters: trivial testlist=sanity-quota
Signed-off-by: Hongchao Zhang <hongchao@whamcloud.com>
Change-Id: I8c34f5f045637e2f44f8cd6806b848ed93d8e784
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56628
Tested-by: Maloo <maloo@whamcloud.com>
Tested-by: jenkins <devops@whamcloud.com>
Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
Reviewed-by: Qian Yingjin <qian@ddn.com>
Reviewed-by: Sergey Cheremencev <scherementsev@ddn.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Oleg Drokin [Fri, 11 Oct 2024 20:40:43 +0000 (16:40 -0400)]
New RC 2.16.0-RC2
Change-Id: I27c5cb17b9a340af7e0cade62c3468f984e7a7e2
Signed-off-by: Oleg Drokin <green@whamcloud.com>
Jian Yu [Fri, 11 Oct 2024 03:00:51 +0000 (20:00 -0700)]
LU-18328 tests: fix old MDS version for sanity-sec/51
Fix the old MDS version for sanity-sec test 51 to skip the test
for symbolic capability names.
The patch also reverts the NODEMAP_TESTS change from commit
v2_16_0-RC1-11-g2a5e8e355498, which casued the following failure:
EXCEPT_7=true: command not found
Test-Parameters: trivial testlist=sanity-sec env=ONLY=51 serverversion=2.15.5
Change-Id: Idf458ad5df7a7d6105b1f4da3e06280cf66919c1
Fixes:
0b0e9162e2 ("LU-13791 mdt: allow using symbolic capability names")
Fixes:
2a5e8e3554 ("LU-18247 nodemap: initialize unused fields on disk")
Signed-off-by: Jian Yu <yujian@whamcloud.com>
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56624
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
Reviewed-by: Sebastien Buisson <sbuisson@ddn.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Jian Yu [Wed, 9 Oct 2024 05:27:45 +0000 (22:27 -0700)]
LU-18331 tests: skip sanityn/33e for older MDS
Skip sanityn test 33e for MDS < 2.15.56.127 since the test depends on
changes made to the MDS code on the server.
Test-Parameters: trivial testlist=sanityn env=ONLY=33e serverversion=2.15.5
Change-Id: I278af93165e928c03436489d90d72a766f27bdd7
Fixes:
2a78a9e2cd ("LU-15527 dne: refactor commit-on-sharing for DNE")
Signed-off-by: Jian Yu <yujian@whamcloud.com>
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56622
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
Reviewed-by: Li Xi <lixi@ddn.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Frederick Dilger [Thu, 3 Oct 2024 18:37:55 +0000 (12:37 -0600)]
LU-15840 tests: interop check for sanity-hsm 262
sanity-hsm.sh test 262 was failing in interop testing with an old
MDS.
Added an interop check to skip the test unless MDS version >=
v2_15_61-204-g5ee13823a4.
Test-Parameters: trivial testlist=sanity-hsm env=ONLY=262
Test-Parameters: testlist=sanity-hsm env=ONLY=262 serverversion=2.15
Signed-off-by: Frederick Dilger <fdilger@whamcloud.com>
Change-Id: I3fc912ad633f7450ae97d9be50627b462d015dea
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56577
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
Reviewed-by: Li Xi <lixi@ddn.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
git://git.whamcloud.com / fs / lustre-release.git / log