When scrubbing a raidz/draid pool, which contains a replacing or
sparing mirror with multiple online children, only one child will
be read. This is not normally a serious concern because the DTL
records are used to determine where a good copy of the data is.
As long as the data can be read from one child the mirror vdev
will use it to repair gaps in any of its children. Furthermore,
even if the data which was read is corrupt the raidz code will
detect this and issue its own repair I/O to correct the damage
in the mirror vdev.
However, in the scenario where the DTL is wrong due to silent
data corruption (say due to overwriting one child) and the scrub
happens to read from a child with good data, then the other damaged
mirror child will not be detected nor repaired.
While this is possible for both raidz and draid vdevs, it's most
pronounced when using draid. This is because by default the zed
will sequentially rebuild a draid pool to a distributed spare,
and the distributed spare half of the mirror is always preferred
since it delivers better performance. This means the damaged
half of the mirror will go undetected even after scrubbing.
For system administrations this behavior is non-intuitive and in
a worst case scenario could result in the only good copy of the
data being unknowingly detached from the mirror.
This change resolves the issue by reading all replacing/sparing
mirror children when scrubbing. When the BP isn't available for
verification, then compare the data buffers from each child. They
must all be identical, if not there's silent damage and an error
is returned to prompt the top-level vdev to issue a repair I/O to
rewrite the data on all of the mirror children. Since we can't
tell which child was wrong a checksum error is logged against the
replacing or sparing mirror vdev.
Reviewed-by: Mark Maybee <mark.maybee@delphix.com>
Reviewed-by: Tony Hutter <hutter2@llnl.gov>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#13555
Commit 63b18e4 fixed an issue in zpl_aio_write() to make sure that
kiocb->ki_pos was updated correctly when opening a file with O_APPEND.
Adding a test to verify O_APPEND functionality with lseek can make
sure that all other distros/kernel versions also have the correct
behavior.
Also moved the threadappends_001_pos test into this append test
directory in functional ZTS directory. This way the two append tests
are together for organization purposes.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Brian Atkinson <batkinson@lanl.gov>
Closes#13424
Only down to tests/zfs-tests/tests, but pull out C programs into the
main Makefile ‒ this means we get correct dependency tracking for all
programs (and parallelise across them)
dist diff:
-zfs-2.1.99/tests/zfs-tests/tests/stress/
-zfs-2.1.99/tests/zfs-tests/tests/stress/Makefile.am
-zfs-2.1.99/tests/zfs-tests/tests/stress/Makefile.in
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Ahelenia Ziemiańska <nabijaczleweli@nabijaczleweli.xyz>
Closes#13316
No installation diff, dist lost
-zfs-2.1.99/cmd/fsck_zfs/fsck.zfs
which was distributed erroneously, since it's generated
Also clean gitrev on clean
Also add -e 'any possible bashisms' to default checkbashisms flags,
and fully parallelise it and shellcheck, and it works out-of-tree, too
Also align the Release in the dist META file correctly
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Ahelenia Ziemiańska <nabijaczleweli@nabijaczleweli.xyz>
Closes#13316
Page writebacks with WB_SYNC_NONE can take several seconds to complete
since they wait for the transaction group to close before being
committed. This is usually not a problem since the caller does not
need to wait. However, if we're simultaneously doing a writeback
with WB_SYNC_ALL (e.g via msync), the latter can block for several
seconds (up to zfs_txg_timeout) due to the active WB_SYNC_NONE
writeback since it needs to wait for the transaction to complete
and the PG_writeback bit to be cleared.
This commit deals with 2 cases:
- No page writeback is active. A WB_SYNC_ALL page writeback starts
and even completes. But when it's about to check if the PG_writeback
bit has been cleared, another writeback with WB_SYNC_NONE starts.
The sync page writeback ends up waiting for the non-sync page
writeback to complete.
- A page writeback with WB_SYNC_NONE is already active when a
WB_SYNC_ALL writeback starts. The WB_SYNC_ALL writeback ends up
waiting for the WB_SYNC_NONE writeback.
The fix works by carefully keeping track of active sync/non-sync
writebacks and committing when beneficial.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Shaan Nobee <sniper111@gmail.com>
Closes#12662Closes#12790
The auto_spare_multiple.ksh test may incorrectly fail for a similar
reason as the auto_spare_shared.ksh test. Add it to known list of
exceptions which should be retried to prevent failures in the CI.
Reviewed-by: George Melikov <mail@gmelikov.ru>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#13318
The redundancy_draid_spare1.ksh and redundancy_draid_spare3.ksh test
cases are a little to strict for the sequential resilver case. While
unlikely it is possible that a handful of correctable checksum errors
will be reported resulting in a test failure. Update the zts-report.py
script to allow this the test case to be retried if requested.
Reviewed-by: George Melikov <mail@gmelikov.ru>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#13318
As found by
git -C tests/ grep ^function | grep -vFe '.lua:' -e '.zcp:' | while IFS=":$IFS" read -r _ _ fn _; do [ $(git -C tests/ grep -wF $fn | head -2 | wc -l) -eq 1 ] && echo $fn; done
after all rounds this comes out to, sorted:
check_slog_state
chgusr_exec
cksum_files
cleanup_pools
compare_modes
count_ACE
dataset_set_defaultproperties
ds_is_snapshot
get_ACE
get_group
get_min
get_mode
get_owner
get_rand_checksum
get_rand_checksum_any
get_rand_large_recsize
get_rand_recsize
get_user_group
getitem
indirect_vdev_mapping_size
is_dilos
log_noresult
log_notinuse
log_other
log_timed_out
log_uninitiated
log_warning
num_jobs_by_cpu
plus_sign_check_l
plus_sign_check_v
record_cksum
rwx_node
seconds_mmp_waits_for_activity
set_cur_usr
setup_mirrors
setup_raidzs
showshares_smb
zfs_zones_setup
This, of course, doesn't catch recursive ones, or ones that log with
their own function name as a prefix, but
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: John Kennedy <john.kennedy@delphix.com>
Reviewed-by: Ryan Moeller <ryan@iXsystems.com>
Signed-off-by: Ahelenia Ziemiańska <nabijaczleweli@nabijaczleweli.xyz>
Closes#13259
This also fixes line welding in test error output
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: John Kennedy <john.kennedy@delphix.com>
Reviewed-by: Ryan Moeller <ryan@iXsystems.com>
Signed-off-by: Ahelenia Ziemiańska <nabijaczleweli@nabijaczleweli.xyz>
Closes#13259
Add a -K option to the test suite to log each test name to /dev/kmsg
(on Linux), so if there's a kernel warning we'll be able to match
it up to a particular test.
Reviewed-by: John Kennedy <john.kennedy@delphix.com>
Signed-off-by: Tony Hutter <hutter2@llnl.gov>
Closes #13227
As previously noted in #12272 the receive-o-x_props_override.ksh test
reliably fails on FreeBSD. Since we don't expect this test to pass
move the exception from the "maybe" to "known" section. This way we
don't retry the FAILED test when it is not expected to pass.
Reviewed-by: George Melikov <mail@gmelikov.ru>
Reviewed-by: Ryan Moeller <freqlabs@FreeBSD.org>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#13167
- Kmemleak `clear` is invoked right before every test case run.
- Kmemleak `scan` is requested right after each test case is finished.
- Kmemleak instrumentation is not used for
setup/cleanup/pretest/posttest/failsafe stages to shorten the test
case execution time.
- Kmemleak periodic scan is disabled (`scan=0`) before the test suite
run to avoid interfering with the on-demand scan results.
- There are unavoidable potential false positives coming from kernel
areas other than OpenZFS module.
- The ZTS with kmemleak enabled duration is increased by ~50%.
Example run
```
Running Time: 07:12:13
Percent passed: 98.3%
unreferenced object 0xffff9da82aea5410 (size 80):
comm "kworker/u32:10", pid 942206, jiffies 4296749716 (age 2615.516s)
hex dump (first 32 bytes):
00 30 30 00 00 00 00 00 ff 8f 30 00 00 00 00 00 .00.......0.....
51 e6 77 05 a8 9d ff ff 00 00 00 00 00 00 00 00 Q.w.............
backtrace:
[<000000005cf1fea2>] alloc_extent_state+0x1d/0xb0 [btrfs]
[<0000000083f78ae5>] set_extent_bit+0x2ff/0x670 [btrfs]
[<00000000de29249e>] lock_extent_bits+0x6b/0xa0 [btrfs]
[<00000000b241f424>] lock_and_cleanup_extent_if_need+0xaf/0x1c0
[btrfs]
[<0000000093ca72b5>] btrfs_buffered_write+0x297/0x7d0 [btrfs]
[<000000002c2938c8>] btrfs_file_write_iter+0x127/0x390 [btrfs]
[<00000000b888f720>] do_iter_readv_writev+0x152/0x1b0
[<00000000320f0bcc>] do_iter_write+0x7c/0x1c0
[<000000000b5a8fe0>] lo_write_bvec+0x62/0x150 [loop]
[<000000009aa03c73>] loop_process_work+0x250/0xbd0 [loop]
[<00000000c7487d8a>] process_one_work+0x1f1/0x390
[<000000000b236831>] worker_thread+0x53/0x3e0
[<0000000023cb3e57>] kthread+0x127/0x150
[<000000002d48676a>] ret_from_fork+0x22/0x30
```
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Ryan Moeller <ryan@iXsystems.com>
Signed-off-by: szubersk <szuberskidamian@gmail.com>
Closes#13084
As explained by the disclaimer in the test case,
"This test can fail since nothing guarantees that old
MOS blocks aren't overwritten."
This behavior is expected and correct, but results in a
flaky test case which is problematic for the CI. The best
we can do to resolve this is to retry the sub-test which
failed when the MOS blocks have clearly been overwritten.
When testing failures were rare enough that a single retry
should normally be sufficient. However, we allow up to
five for good measure.
Reviewed by: George Melikov <mail@gmelikov.ru>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#13119
When attaching a vdev to a mirror wait for the resilver to complete
before invoking `zdb` to inspect the pool. This ensures the pool is
essentially idle which allows `zdb` to open the imported pool reliably.
Reviewed-by: John Kennedy <john.kennedy@delphix.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#13112Closes#6935
The dRAID section of the zpool_expand_001_pos test would reliably fail
because the calculated expansion size assumed the dRAID top-level vdev
was created with a distributed spare. Create the vdev as expected to
resolve the test failure.
This test case flaw was accidentally caused by changing the default
number of dRAID distributed spares from one to zero while dRAID was
being developed.
Additionally, remove zpool_expand_005_pos from the list of possible
faulty tests. It appears to be passing consistently in my testing.
Reviewed by: George Melikov <mail@gmelikov.ru>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#13091
The on-disk cost of creating a snapshot or bookmark is sufficiently low
that it is difficult to make it reliably fail even when the pool is
"full". In order to avoid false positives remove these two checks from
the test case.
Reviewed-by: George Melikov <mail@gmelikov.ru>
Reviewed-by: John Kennedy <john.kennedy@delphix.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#13060
CLOCK_MONOTONIC_RAW is only a thing on Linux and macOS. I'm not
actually sure why the previous hardcoding of a constant didn't
error out, but when we removed it, it sure does now.
Reviewed-by: Alexander Motin <mav@FreeBSD.org>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Co-authored-by: Rich Ercolani <rincebrain@gmail.com>
Signed-off-by: Rich Ercolani <rincebrain@gmail.com>
Closes#12995
Deprecation of Python versions below 3.6 gives opportunity to unify the
build and install requirements for OpenZFS packages. The minimal
supported Python version is 3.6 as this is the most recent Python
package CentOS/RHEL 7 users can get.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Rich Ercolani <rincebrain@gmail.com>
Reviewed-by: John Kennedy <john.kennedy@delphix.com>
Signed-off-by: szubersk <szuberskidamian@gmail.com>
Closes#12925
Under Linux when rolling back a mounted filesystem negative dentries
may not be dropped from the cache. This can result in an ENOENT
being incorrectly returned on first access. Issuing a `df` before
the unmount results in the negative dentries being invalidated and
side steps the issue.
This is solely a workaround for the test case on Linux and not
correct behavior. The core issue of invalidating negative dentries
needs to be handled with a kernel side change. This is being
tracked as issue #6143.
Reviewed-by: George Melikov <mail@gmelikov.ru>
Reviewed-by: John Kennedy <john.kennedy@delphix.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#12898
Issue #6143
The rerefreserv_raidz test was failing on Linux because the sync being
issued doesn't guarantee a pool sync. Switch to using the sync_pool
function and remove the ZTS exception for Linux.
Reviewed-by: George Melikov <mail@gmelikov.ru>
Reviewed-by: John Kennedy <john.kennedy@delphix.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#12897
The rsend_007_pos test reliably fails on Linux in the cleanup
function. This is caused by an unmount error when attempting to
recursively destroy the newly received datasets. Invoking `df`
prior to the `zfs destroy` interestingly avoids the unmont error.
Why this should matter is unclear and should be investigated.
However, this minor tweak may allow us to remove the ZTS rsend
exceptions. The subsequent rsend_010_pos and rsend_011_pos
failures were a result of this initial failure. The other
"maybe" failures I was unable to reproduce and have not been
recently observed in the master branch.
Reviewed-by: Tony Nguyen <tony.nguyen@delphix.com>
Reviewed-by: Ryan Moeller <ryan@iXsystems.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#5665Closes#6086Closes#6087Closes#6446Closes#12876
The alloc_class_* tests may fail on Linux with an EBUSY error if
`zfs destroy` is run before the `dd` process has had a chance to
terminate. Wait on the pid after the `kill -9` to make sure.
When testing I didn't observe any failures for the alloc_class
tests. Remove them from the exceptions list, the CI was used to
verify the tests pass on all platforms.
Reviewed-by: John Kennedy <john.kennedy@delphix.com>
Reviewed-by: Rich Ercolani <rincebrain@gmail.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#12873
The import_rewind_device_replaced.ksh test was never entirely reliable
because it depends on MOS data not being overwritten. The MOS data is
not protected by the snapshot so occasional failures were always
expected. However, this test is now failing reliably on all platforms
indicating something has changed in the code since the test was marked
"maybe". Convert the test to a "known" failure until the root cause
is identified and resolved.
Reviewed-by: John Kennedy <john.kennedy@delphix.com>
Reviewed-by: George Melikov <mail@gmelikov.ru>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#12821
Due to a possible lock inversion the zvol open call path on Linux
needs to be able to retry in the case where the spa_namespace_lock
cannot be acquired.
For Linux 5.12 an older kernel this was accomplished by returning
-ERESTARTSYS from zvol_open() to request that blkdev_get() drop
the bdev->bd_mutex lock, reaquire it, then call the open callback
again. However, as of the 5.13 kernel this behavior was removed.
Therefore, for 5.12 and older kernels we preserved the existing
retry logic, but for 5.13 and newer kernels we retry internally in
zvol_open(). This should always succeed except in the case where
a pool's vdev are layed on zvols, in which case it may fail. To
handle this case vdev_disk_open() has been updated to retry when
opening a device when -ERESTARTSYS is returned.
Reviewed-by: Tony Hutter <hutter2@llnl.gov>
Reviewed-by: Tony Nguyen <tony.nguyen@delphix.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Issue #12301Closes#12759
This was a project proposed as part of the Quality theme for the
hackthon for the 2021 OpenZFS Developer Summit. The idea is to improve
the usability of the automated tests that get run when a PR is created
by having failing tests automatically rerun in order to make flaky
tests less impactful.
Reviewed-by: John Kennedy <john.kennedy@delphix.com>
Reviewed-by: Tony Nguyen <tony.nguyen@delphix.com>
Signed-off-by: Paul Dagnelie <pcd@delphix.com>
Closes#12740
This test case may fail on 5.13 and newer Linux kernels if the
/dev/zvol/ device is not created by udev.
Reviewed-by: Rich Ercolani <rincebrain@gmail.com>
Reviewed-by: John Kennedy <john.kennedy@delphix.com>
Reviewed-by: George Melikov <mail@gmelikov.ru>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Issue #12301
Closes #12738
It keeps failing, on changes which aren't related at all.
So until someone runs down why, I'd like it to stop being the
sole reason for CI failures.
Reviewed-by: John Kennedy <john.kennedy@delphix.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: George Melikov <mail@gmelikov.ru>
Signed-off-by: Rich Ercolani <rincebrain@gmail.com>
Closes#12733
The intention of the zfs_iter_mounted() is to traverse the dataset
and its descendants, not the snapshots. The current code can cause
a mounted snapshot to be included and thus zfs_open() on the snapshot
with ZFS_TYPE_FILESYSTEM would print confusing message such as "cannot
open 'rpool/fs@snap': snapshot delimiter '@' is not expected here".
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Youzhong Yang <yyang@mathworks.com>
Closes#12447Closes#12448
Add the following test failures to the exception list for FreeBSD
to ensure we notice new unexpected failures.
pool_checkpoint/checkpoint_big_rewind
pool_checkpoint/checkpoint_indirect
And the following for Linux.
zvol/zvol_misc/zvol_misc_snapdev
Reviewed-by: George Melikov <mail@gmelikov.ru>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Issue #12621
Issue #12622
Issue #12623Closes#12624
The zvol_misc tests, in particular zvol_misc_volmode, make use of a
common udev_wait function to wait for zvol devices in /dev to quiesce
on Linux. On other platforms this function currently only sleeps for
one second before returning. This is insufficient, and
zvol_misc_volmode has been flaky on FreeBSD as a result.
Replace udev_wait with block_device_wait, passing through the optional
device parameter where possible. Rearrange a few checks to strengthen
the verifications we are making and avoid unnecessarily sleeping. We
must keep udev_wait in a couple places to pass in Github CI workflows.
Remove zvol_misc_volmode from the maybe failing tests on FreeBSD in
zts-report.py.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: John Kennedy <john.kennedy@delphix.com>
Signed-off-by: Ryan Moeller <ryan@iXsystems.com>
Closes#12583
Issue #11854 has been resolved, so we can remove the exceptions for it.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: John Kennedy <john.kennedy@delphix.com>
Reviewed-by: George Melikov <mail@gmelikov.ru>
Signed-off-by: Ryan Moeller <ryan@iXsystems.com>
Closes#12527
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Alexander Motin <mav@FreeBSD.org>
Reviewed-by: Ryan Moeller <ryan@iXsystems.com>
Signed-off-by: Ka Ho Ng <khng@FreeBSD.org>
Sponsored-by: The FreeBSD Foundation
Closes#12458
Reviewed-by: Tony Nguyen <tony.nguyen@delphix.com>
Reviewed-by: Allan Jude <allan@klarasystems.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Alexander Motin <mav@FreeBSD.org>
Signed-off-by: Ryan Moeller <ryan@iXsystems.com>
Closes#12432
In l2arc_add_vdev() first decide whether the device is eligible for
L2ARC rebuild or whole device trim and then add it to the list of cache
devices. Otherwise l2arc_feed_thread() might already start writing on
the device invalidating previous content as l2ad_hand = l2ad_start.
However l2arc_rebuild_vdev() needs the device present in the cache
device list to figure out its l2arc_dev_t. Fix this by moving most of
l2arc_rebuild_vdev() in a new function l2arc_rebuild_dev() which does
not need to search in the cache device list.
In contrast to l2arc_add_vdev() we do not have to worry about
l2arc_feed_thread() invalidating previous content when onlining a
cache device. The device parameters (l2ad*) are not cleared when
offlining the device and writing new buffers will not invalidate
all previous content. In worst case only buffers that have not had
their log block written to the device will be lost.
Retire persist_l2arc_00{4,5,8} tests since they cover code already
covered by the remaining ones. Test persist_l2arc_006 is renamed to
persist_l2arc_004 and persist_l2arc_007 is renamed to persist_l2arc_005.
Fix a typo in persist_l2arc_004, and remove an assertion that is not
always true from l2arc_arcstats_pos. Also update an assertion in
persist_l2arc_005 and explain why in a comment.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: George Amanakis <gamanakis@gmail.com>
Closes#12365
This enables ZED to auto-online vdevs that are not wholedisk managed by
ZFS.
Signed-off-by: Ryan Moeller <ryan@iXsystems.com>
Reviewed-by: Don Brady <don.brady@delphix.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Tony Hutter <hutter2@llnl.gov>
The receive-o-x_props_override test case reliably fails on the
FreeBSD main builders (but not on Linux), until the root cause is
understood add this test to the FreeBSD exception list.
On Linux the alloc_class_012_pos test case may occasionally fail.
This is a known false positive which has also been added to the
Linux exception list until the test can be made entirely reliable.
Reviewed-by: George Melikov <mail@gmelikov.ru>
Reviewed-by: John Kennedy <john.kennedy@delphix.com>
Reviewed-by: Ryan Moeller <ryan@iXsystems.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#12272
