This commit partially reverts changes to multilists in PR 7968
(multi-threaded spa-sync()) and adds some cache line alignments to
separate read-only multilists and heavily modified refcount's to
different cache lines.
Reviewed-by: Matthew Ahrens <mahrens@delphix.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Alexander Motin <mav@FreeBSD.org>
Sponsored-by: iXsystems, Inc.
Closes#12158
This mostly reverts "3537 want pool io kstats" commit of 8 years ago.
From one side this code using pool-wide locks became pretty bad for
performance, creating significant lock contention in I/O pipeline.
From another, there are more efficient ways now to obtain detailed
statistics, while this statistics is illumos-specific and much less
usable on Linux and FreeBSD, reported only via procfs/sysctls.
This commit does not remove KSTAT_TYPE_IO implementation, that may
be removed later together with already unused KSTAT_TYPE_INTR and
KSTAT_TYPE_TIMER.
Reviewed-by: Matthew Ahrens <mahrens@delphix.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Alexander Motin <mav@FreeBSD.org>
Sponsored-By: iXsystems, Inc.
Closes#12212
`getfsstat(2)` is used to retrieve the list of mounted file systems,
which libzfs uses when fetching properties like mountpoint, atime,
setuid, etc. The `mode` parameter may be `MNT_NOWAIT`, which uses
information in the VFS's cache, or `MNT_WAIT`, which effectively does a
`statfs` on every single mounted file system in order to fetch the most
up-to-date information. As far as I can tell, the only fields that
libzfs cares about are the filesystem's name, mountpoint, fstypename,
and mount flags. Those things are always updated on mount and unmount,
so they will always be accurate in the VFS's mount cache except in two
circumstances:
1) When a file system is busy unmounting
2) When a ZFS file system changes the value of a mount-overridable
property like atime or setuid, but doesn't remount the file system.
Right now that only happens when the property is changed by an
unprivileged user who has delegated authority to change the property
but not to mount the dataset. But perhaps libzfs could choose to do
it for other reasons in the future.
Switching to `MNT_NOWAIT` will greatly improve speed with no downside,
as long as we explicitly update the mount cache whenever we change a
mount-overridable property.
For comparison, Illumos gets this information using the native
`getmntany` and `getmntent` functions, which also use cached
information. The illumos function that would refresh the cache,
`resetmnttab`, is never called by libzfs.
And on GNU/Linux, `getmntany` and `getmntent` don't even communicate
with the kernel directly. They simply parse the file they are given,
which is usually /etc/mtab or /proc/mounts. Perhaps the implementation
of /proc/mounts is synchronous, ala MNT_WAIT; I don't know.
Sponsored-by: Axcient
Reviewed-by: Ryan Moeller <ryan@iXsystems.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Alan Somers <asomers@gmail.com>
Closes: #12091
Update the logic to handle the dedup-case of consecutive
FREEs in the livelist code. The logic still ensures that
all the FREE entries are matched up with a respective
ALLOC by keeping a refcount for each FREE blkptr that we
encounter and ensuring that this refcount gets to zero
by the time we are done processing the livelist.
zdb -y no longer panics when encountering double frees
Reviewed-by: Matthew Ahrens <mahrens@delphix.com>
Reviewed-by: John Kennedy <john.kennedy@delphix.com>
Reviewed-by: Don Brady <don.brady@delphix.com>
Signed-off-by: Serapheim Dimitropoulos <serapheim@delphix.com>
Closes#11480Closes#12177
- Avoid atomic_add() when updating as_lower_bound/as_upper_bound.
Previous code was excessively strong on 64bit systems while not
strong enough on 32bit ones. Instead introduce and use real
atomic_load() and atomic_store() operations, just an assignments
on 64bit machines, but using proper atomics on 32bit ones to avoid
torn reads/writes.
- Reduce number of buckets on large systems. Extra buckets not as
much improve add speed, as hurt reads. Unlike wmsum for aggsum
reads are still important.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Ryan Moeller <ryan@iXsystems.com>
Signed-off-by: Alexander Motin <mav@FreeBSD.org>
Sponsored-By: iXsystems, Inc.
Closes#12145
wmsum counters are a reduced version of aggsum counters, optimized for
write-mostly scenarios. They do not provide optimized read functions,
but instead allow much cheaper add function. The primary usage is
infrequently read statistic counters, not requiring exact precision.
The Linux implementation is directly mapped into percpu_counter KPI.
The FreeBSD implementation is directly mapped into counter(9) KPI.
In user-space due to lack of better implementation mapped to aggsum.
Unfortunately neither Linux percpu_counter nor FreeBSD counter(9)
provide sufficient functionality to completelly replace aggsum, so
it still remains to be used for several hot counters.
Reviewed-by: Paul Dagnelie <pcd@delphix.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Ryan Moeller <ryan@iXsystems.com>
Signed-off-by: Alexander Motin <mav@FreeBSD.org>
Sponsored-By: iXsystems, Inc.
Closes#12114
Previously, ZFS scaled maxinflight_bytes based on total number of
disks in the pool. A 3-wide mirror was receiving a queue depth of 3
disks, which it should not, since it reads from all the disks inside.
For wide raidz the situation was slightly better, but still a 3-wide
raidz1 received a depth of 3 disks instead of 2.
The new code counts only unique data disks, i.e. 1 disk for mirrors
and non-parity disks for raidz/draid. For draid the math is still
imperfect, since vdev_get_nparity() returns number of parity disks
per group, not per vdev, but still some better than it was.
This should slightly reduce scrub influence on payload for some pool
topologies by avoiding excessive queuing.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Ryan Moeller <ryan@iXsystems.com>
Signed-off-by: Alexander Motin <mav@FreeBSD.org>
Sponsored-By: iXsystems, Inc.
Closing #12046
Propagate vdev child state to parents on invalid label
Add VDEV_AUX_BAD_LABEL to print_import_config()
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Mark Maybee <mark.maybee@delphix.com>
Co-authored-by: Srikanth N S <srikanth.nagasubbaraoseetharaman@hpe.com>
Signed-off-by: Vipin Kumar Verma <vipin.verma@hpe.com>
Closes#12088
This change addresses two distinct scenarios which are possible
when performing a sequential resilver to a dRAID pool with vdevs
that contain silent unknown damage. Which in this circumstance
took the form of the devices being intentionally overwritten with
zeros. However, it could also result from a device returning incorrect
data while a sequential resilver was in progress.
Scenario 1) A sequential resilver is performed while all of the
dRAID vdevs are ONLINE and there is silent damage present on the
vdev being resilvered. In this case, nothing will be repaired
by vdev_raidz_io_done_reconstruct_known_missing() because
rc->rc_error isn't set on any of the raid columns. To address
this vdev_draid_io_start_read() has been updated to always mark
the resilvering column as ESTALE for sequential resilver IO.
Scenario 2) Multiple columns contain silent damage for the same
block and a sequential resilver is performed. In this case it's
impossible to generate the correct data from parity unless all of
the damaged columns are being sequentially resilvered (and thus
only good data is used to generate parity). This is as expected
and there's nothing which can be done about it. However, we need
to be careful not to make to situation worse. Since we can't
verify the data is actually good without a checksum, we must
only repair the devices which are being sequentially resilvered.
Otherwise, an incorrect repair to a device which previously
contained good data could effectively lock in the damage and
make reconstruction impossible. A check for this was added to
vdev_raidz_io_done_verified() along with a new test case.
Lastly, this change updates the redundancy_draid_spare1 and
redundancy_draid_spare3 test cases to be more representative
of normal dRAID replacement operation. Specifically, what we
care about is that the scrub run after a sequential resilver
does not find additional blocks which need repair. This would
indicate the sequential resilver failed to rebuild a section of
one of the devices. Note also the tests were switched to using
the verify_pool() function which still checks for checksum errors.
Reviewed-by: Mark Maybee <mark.maybee@delphix.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#12061
While use of dynamic taskqs allows to reduce number of idle threads,
hardcoded 8 taskqs of each kind is a big overkill for small systems,
complicating CPU scheduling, increasing I/O reorder, etc, while
providing no real locking benefits, just not needed there.
On another side, 12*8 worker threads per kind are able to overload
almost any system nowadays. For example, pool of several fast SSDs
with SHA256 checksum makes system barely responsive during scrub, or
with dedup enabled barely responsive during large file deletion.
To address both problems this patch introduces ZTI_SCALE macro, alike
to ZTI_BATCH, but with multiple taskqs, depending on number of CPUs,
to be used in places where lock scalability is needed, while request
ordering is not so much. The code is made to create new taskq for
~6 worker threads (less for small systems, but more for very large)
up to 80% of CPU cores (previous 75% was not good for rounding down).
Both number of threads and threads per taskq are now tunable in case
somebody really wants to use all of system power for ZFS.
While obviously some benchmarks show small peak performance reduction
(not so big really, especially on systems with SMT, where use of the
second threads does not give as much performance as the first ones),
they also show dramatic latency reduction and much more smooth user-
space operation in case of high CPU usage by ZFS.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Alexander Motin <mav@FreeBSD.org>
Sponsored-By: iXsystems, Inc.
Closes#11966
Use dsl_dataset_has_resume_receive_state()
not dsl_dataset_is_zapified() to check if
stream is resumable.
Reviewed-by: Matthew Ahrens <mahrens@delphix.com>
Reviewed-by: Alek Pinchuk <apinchuk@axcient.com>
Reviewed-by: Ryan Moeller <ryan@ixsystems.com>
Signed-off-by: Paul Zuchowski <pzuchowski@datto.com>
Closes#12034
Commit d1d4769 takes into account the encryption key version to
decide if the local_mac could be zeroed out. However, this could lead
to failure mounting encrypted datasets created with intermediate
versions of ZFS encryption available in master between major releases.
In order to prevent this situation revert d1d4769 pending a more
comprehensive fix which addresses the mount failure case.
Reviewed-by: George Amanakis <gamanakis@gmail.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Issue #11294
Issue #12025
Issue #12300Closes#12033
zfs_zevent_console committed multiple printk()s per line without
properly continuing them ‒ a single event could easily be fragmented
across over thirty lines, making it useless for direct application
zfs_zevent_cols exists purely to wrap the output from zfs_zevent_console
The niche this was supposed to fill can be better served by something
akin to the all-syslog ZEDLET
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Ahelenia Ziemiańska <nabijaczleweli@nabijaczleweli.xyz>
Closes#7082Closes#11996
When dRAID performs a normal read operation only the data columns
in the raid map are read from disk. This is enough information to
calculate the checksum, verify it, and return the needed data to the
application. It's only in the event of a checksum failure that the
additional parity and any empty columns must be read since they are
required for parity reconstruction.
Reading these additional columns is handled by vdev_raidz_read_all()
which calls vdev_draid_map_alloc_empty() to expand the raid_map_t
and submit IOs for the missing columns. This all works correctly,
but it fails to account for any "short" columns. These are data
columns which are padded with a empty skip sector at the end.
Since that empty sector is not needed for a normal read it's not
read when columns is first read from disk. However, like the parity
and empty columns the skip sector is needed to perform reconstruction.
The fix is to mark any "short" columns as never being read by clearing
the rc_tried flag when expanding the raid_map_t. This will cause
the entire column to re-read from disk in the event of a checksum
failure allowing the self-healing functionality to repair the block.
Note that this only effects the self-healing feature because when
scrubbing a pool the parity, data, and empty columns are all read
initially to verify their contents. Furthermore, only blocks which
contain "short" columns would be effected, and only when the memory
backing the skip sector wasn't already zeroed out.
This change extends the existing redundancy_raidz.ksh test case to
verify self-healing (as well as resilver and scrub). Then applies
the same test case to dRAID with a slightly modified version of
the test script called redundancy_draid.ksh. The unused variable
combrec was also removed from both test cases.
Reviewed-by: Matthew Ahrens <mahrens@delphix.com>
Reviewed-by: Mark Maybee <mark.maybee@delphix.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#12010
Previous code tried to keep prefetch streams while moving dnode. But
it was at least not updating per-stream zs_fetchback pointers, causing
use-after-free on next access. Instead of that I see much easier and
cleaner to just drop old prefetch state and start new from scratch.
Reviewed-by: Matthew Ahrens <mahrens@delphix.com>
Reviewed-by: Igor Kozhukhov <igor@dilos.org>
Signed-off-by: Alexander Motin <mav@FreeBSD.org>
Sponsored-By: iXsystems, Inc.
Closes#11936Closes#11998
This ensures that we don't accumulate checksum errors against offline or
unavailable devices but, more importantly, means that we don't
needlessly create DTL entries for offline devices that are already
up-to-date.
Consider a 3-way mirror, with disk A always online (and so always with
an empty DTL) and B and C only occasionally online. When A & B resilver
with C offline, B's DTL will effectively be appended to C's due to these
spurious ZIOs even as the resilver empties B's DTL:
* These ZIOs land in vdev_mirror_scrub_done() and flag an error
* That flagged error causes vdev_mirror_io_done() to see
unexpected_errors, so it issues a ZIO_TYPE_WRITE repair ZIO, which
inherits ZIO_FLAG_SCAN_THREAD because zio_vdev_child_io() includes
that flag in ZIO_VDEV_CHILD_FLAGS.
* That ZIO fails, too, and eventually zio_done() gets its hands on it
and calls vdev_stat_update().
* vdev_stat_update() sees the error and this zio...
* is not speculative,
* is not due to EIO (but rather ENXIO, since the device is closed)
* has an ->io_vd != NULL (specifically, the offline leaf device)
* is a write
* is for a txg != 0 (but rather the read block's physical birth txg)
* has ZIO_FLAG_SCAN_THREAD asserted
* So: vdev_stat_update() calls vdev_dtl_dirty() on the offline vdev.
Then, when A & C resilver with B offline, that story gets replayed and
C's DTL will be appended to B's.
In fact, one does not need this permanently-broken-mirror scenario to
induce badness: breaking a mirror with no DTLs and then scrubbing will
create DTLs for all offline devices. These DTLs will persist until the
entire mirror is reassembled for the duration of the *resilver*, which,
incidentally, will not consider the devices with good data to be sources
of good data in the case of a read failure.
Reviewed-by: Mark Maybee <mark.maybee@delphix.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Nathaniel Wesley Filardo <nwfilardo@gmail.com>
Closes#11930
This deduplicates 2 sets of caches which use the same allocation size.
Memory savings fluctuate a lot, one sample result is FreeBSD running
"make buildworld" saving ~180MB RAM in reduced page count associated
with zio caches.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Mateusz Guzik <mjguzik@gmail.com>
Closes#11877
Fix NULL pointer dereference when reporting
checksum error for gang block in zio_done.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Paul Zuchowski <pzuchowski@datto.com>
Closes#11872Closes#11896
Traversal code, traverse_visitbp() does visit blocks recursively.
Indirect (Non L0) Block of size 128k could contain, 1024 block pointers
of 128 bytes. In case of full traverse OR incremental traverse, where
all blocks were modified, it could traverse large number of blocks
pointed by indirect. Traversal code does issue prefetch of blocks
traversed below indirect. This could result into large number of
async reads queued on vdev queue. So, account for prefetch issued for
blocks pointed by indirect and limit max prefetch in one go.
Module Param:
zfs_traverse_indirect_prefetch_limit: Limit of prefetch while traversing
an indirect block.
Local counters:
prefetched: Local counter to account for number prefetch done.
pidx: Index for which next prefetch to be issued.
ptidx: Index at which next prefetch to be triggered.
Keep "ptidx" somewhere in the middle of blocks prefetched, so that
blocks prefetch read gets the enough time window before their demand
read is issued.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Alexander Motin <mav@FreeBSD.org>
Signed-off-by: Jitendra Patidar <jitendra.patidar@nutanix.com>
Closes#11802Closes#11803
It's been observed in the CI that the required 25% of obsolete bytes
in the mapping can be to high a threshold for this test resulting in
condensing never being triggered and a test failure. To prevent these
failures make the existing zfs_condense_indirect_obsolete_pct tuning
available so the obsolete percentage can be reduced from 25% to 5%
during this test.
Reviewed-by: Ryan Moeller <ryan@iXsystems.com>
Reviewed-by: George Melikov <mail@gmelikov.ru>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#11869
When a rebuild completes it will automatically schedule a follow up
scrub to verify all of the block checksums. Before setting up the
scrub execute the counterpart dsl_scan_setup_check() function to
confirm the scrub can be started. Prior to this change we'd only
check vdev_rebuild_active() which isn't as comprehensive, and using
the check function keeps all of this logic in one place.
Reviewed-by: Mark Maybee <mark.maybee@delphix.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#11849
Just as delay zevents can flood the zevent pipe when a vdev becomes
unresponsive, so do the deadman zevents.
Ratelimit deadman zevents according to the same tunable as for delay
zevents.
Enable deadman tests on FreeBSD and add a test for deadman event
ratelimiting.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Don Brady <don.brady@delphix.com>
Signed-off-by: Ryan Moeller <ryan@iXsystems.com>
Closes#11786
Correct an assortment of typos throughout the code base.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Matthew Ahrens <mahrens@delphix.com>
Reviewed-by: Ryan Moeller <ryan@iXsystems.com>
Signed-off-by: Andrea Gelmini <andrea.gelmini@gelma.net>
Closes#11774
The lower bound for this scaling to too low and the upper bound is too
high. Use a fixed default length of 512 instead, which is a reasonable
value on any system.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Ryan Moeller <ryan@iXsystems.com>
Closes#11822
ratelimit_dropped isn't protected by a lock and is expected to
be updated atomically.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Ryan Moeller <ryan@iXsystems.com>
Closes#11822
For gang blocks, `DVA_GET_ASIZE()` is the total space allocated for the
gang DVA including its children BP's. The space allocated at each DVA's
vdev/offset is `vdev_psize_to_asize(vd, SPA_GANGBLOCKSIZE)`.
This commit makes this relationship more clear by using a helper
function, `vdev_gang_header_asize()`, for the space allocated at the
gang block's vdev/offset.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Matthew Ahrens <mahrens@delphix.com>
Closes#11744
Reviewed-by: Matthew Ahrens <mahrens@delphix.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Andrea Gelmini <andrea.gelmini@gelma.net>
Closes#11775
To make better predictions on parallel workloads dmu_zfetch() should
be called as early as possible to reduce possible request reordering.
In particular, it should be called before dmu_buf_hold_array_by_dnode()
calls dbuf_hold(), which may sleep waiting for indirect blocks, waking
up multiple threads same time on completion, that can significantly
reorder the requests, making the stream look like random. But we
should not issue prefetch requests before the on-demand ones, since
they may get to the disks first despite the I/O scheduler, increasing
on-demand request latency.
This patch splits dmu_zfetch() into two functions: dmu_zfetch_prepare()
and dmu_zfetch_run(). The first can be executed as early as needed.
It only updates statistics and makes predictions without issuing any
I/Os. The I/O issuance is handled by dmu_zfetch_run(), which can be
called later when all on-demand I/Os are already issued. It even
tracks the activity of other concurrent threads, issuing the prefetch
only when _all_ on-demand requests are issued.
For many years it was a big problem for storage servers, handling
deeper request queues from their clients, having to either serialize
consequential reads to make ZFS prefetcher usable, or execute the
incoming requests as-is and get almost no prefetch from ZFS, relying
only on deep enough prefetch by the clients. Benefits of those ways
varied, but neither was perfect. With this patch deeper queue
sequential read benchmarks with CrystalDiskMark from Windows via
iSCSI to FreeBSD target show me much better throughput with almost
100% prefetcher hit rate, comparing to almost zero before.
While there, I also removed per-stream zs_lock as useless, completely
covered by parent zf_lock. Also I reused zs_blocks refcount to track
zf_stream linkage of the stream, since I believe previous zs_fetch ==
NULL check in dmu_zfetch_stream_done() was racy.
Delete prefetch streams when they reach ends of files. It saves up
to 1KB of RAM per file, plus reduces searches through the stream list.
Block data prefetch (speculation and indirect block prefetch is still
done since they are cheaper) if all dbufs of the stream are already
in DMU cache. First cache miss immediately fires all the prefetch
that would be done for the stream by that time. It saves some CPU
time if same files within DMU cache capacity are read over and over.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Adam Moss <c@yotes.com>
Reviewed-by: Matthew Ahrens <mahrens@delphix.com>
Signed-off-by: Alexander Motin <mav@FreeBSD.org>
Sponsored-By: iXsystems, Inc.
Closes#11652
If TX_WRITE is create on a file, and the file is later deleted and a new
directory is created on the same object id, it is possible that when
zil_commit happens, zfs_get_data will be called on the new directory.
This may result in panic as it tries to do range lock.
This patch fixes this issue by record the generation number during
zfs_log_write, so zfs_get_data can check if the object is valid.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Chunwei Chen <david.chen@nutanix.com>
Closes#10593Closes#11682
Commit 235a85657 introduced a regression in evaluation of POSIX modes
that require group DENY entries in the internal ZFS ACL. An example
of such a POSX mode is 007. When write_implies_delete_child is set,
then ACE_WRITE_DATA is added to `wanted_dirperms` in prior to calling
zfs_zaccess_common(). This occurs is zfs_zaccess_delete().
Unfortunately, when zfs_zaccess_aces_check hits this particular DENY
ACE, zfs_groupmember() is checked to determine whether access should be
denied, and since zfs_groupmember() always returns B_TRUE on Linux and
so this check is failed, resulting ultimately in EPERM being returned.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Ryan Moeller <ryan@iXsystems.com>
Signed-off-by: Andrew Walker <awalker@ixsystems.com>
Closes#11760
The FreeBSD boot loader relies on the bootfs property and is capable
of booting from removed (indirect) vdevs.
Reviewed-by Eric van Gyzen
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Martin Matuska <mm@FreeBSD.org>
Closes#11763
= Motivation
We've noticed several zloop crashes within Delphix generated
due to the following sequence of events:
- A device gets expanded and new metaslabas are allocated for
it. These metaslabs go through `metaslab_init()` but haven't
gone through `metaslab_sync_done()` yet. This meas that the
only range tree that's actually set is the `ms_allocatable`.
All the others are NULL.
- A vdev_initialization is issues and `vdev_initialize_thread`
starts processing one of these new metaslabs of the expanded
vdev.
- As part of `vdev_initialize_calculate_progress()` we call
into `metaslab_load()` and `metaslab_load_impl()` which
in turn tries to dereference the metaslabs trees that
are still NULL and therefore we crash.
The same failure can come up from the `vdev_trim` code paths.
= This Patch
We considered the following solutions to deal with this issue:
[A] Add logic to `vdev_initialize/trim` to skip those new
metaslabs. We decided against this as it would be good
to avoid exposing this lower-level detail to higer-level
operations.
[B] Have `metaslab_load_impl()` return early for new metaslabs
and thus never touch those range_trees that are NULL at
that time. This seemed more of a work-around for the bug
and not a clear-cut solution.
[C] Refactor our logic so all metaslabs have their range_trees
created at the time of their creatin in `metaslab_init()`.
In this patch we decided to go with [C] because:
(1) It doesn't expose more metaslab details to higher level
operations such as vdev initialize and trim.
(2) The current behavior of creating the range trees lazily
in `metaslab_sync_done()` is unnecessarily complicated.
(3) Always initializing the metaslab range_trees makes other
parts of the codebase cleaner. For example, we used to
use `ms_freed` as the reference value for knowing whether
all the range_trees have been initialized. Now we no
longer need to do that check in most places (and in the
few that we do we use the `ms_new` boolean field now
which is more readable).
= Side Changes
Probably due to a mismerge we set `ms_loaded` to `B_TRUE` twice
in `metasloab_load_impl()`. In this patch we remove the extraneous
assignment.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Matthew Ahrens <mahrens@delphix.com>
Signed-off-by: Serapheim Dimitropoulos <serapheim@delphix.com>
Closes#11737
The RAIDZ and DRAID code is responsible for reporting checksum errors on
their child vdevs. Checksum errors represent events where a disk
returned data or parity that should have been correct, but was not. In
other words, these are instances of silent data corruption. The
checksum errors show up in the vdev stats (and thus `zpool status`'s
CKSUM column), and in the event log (`zpool events`).
Note, this is in contrast with the more common "noisy" errors where a
disk goes offline, in which case ZFS knows that the disk is bad and
doesn't try to read it, or the device returns an error on the requested
read or write operation.
RAIDZ/DRAID generate checksum errors via three code paths:
1. When RAIDZ/DRAID reconstructs a damaged block, checksum errors are
reported on any children whose data was not used during the
reconstruction. This is handled in `raidz_reconstruct()`. This is the
most common type of RAIDZ/DRAID checksum error.
2. When RAIDZ/DRAID is not able to reconstruct a damaged block, that
means that the data has been lost. The zio fails and an error is
returned to the consumer (e.g. the read(2) system call). This would
happen if, for example, three different disks in a RAIDZ2 group are
silently damaged. Since the damage is silent, it isn't possible to know
which three disks are damaged, so a checksum error is reported against
every child that returned data or parity for this read. (For DRAID,
typically only one "group" of children is involved in each io.) This
case is handled in `vdev_raidz_cksum_finish()`. This is the next most
common type of RAIDZ/DRAID checksum error.
3. If RAIDZ/DRAID is not able to reconstruct a damaged block (like in
case 2), but there happens to be additional copies of this block due to
"ditto blocks" (i.e. multiple DVA's in this blkptr_t), and one of those
copies is good, then RAIDZ/DRAID compares each sector of the data or
parity that it retrieved with the good data from the other DVA, and if
they differ then it reports a checksum error on this child. This
differs from case 2 in that the checksum error is reported on only the
subset of children that actually have bad data or parity. This case
happens very rarely, since normally only metadata has ditto blocks. If
the silent damage is extensive, there will be many instances of case 2,
and the pool will likely be unrecoverable.
The code for handling case 3 is considerably more complicated than the
other cases, for two reasons:
1. It needs to run after the main raidz read logic has completed. The
data RAIDZ read needs to be preserved until after the alternate DVA has
been read, which necessitates refcounts and callbacks managed by the
non-raidz-specific zio layer.
2. It's nontrivial to map the sections of data read by RAIDZ to the
correct data. For example, the correct data does not include the parity
information, so the parity must be recalculated based on the correct
data, and then compared to the parity that was read from the RAIDZ
children.
Due to the complexity of case 3, the rareness of hitting it, and the
minimal benefit it provides above case 2, this commit removes the code
for case 3. These types of errors will now be handled the same as case
2, i.e. the checksum error will be reported against all children that
returned data or parity.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Matthew Ahrens <mahrens@delphix.com>
Closes#11735
The `rr_code` field in `raidz_row_t` is unused.
This commit removes the field, as well as the code that's used to set
it.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Matthew Ahrens <mahrens@delphix.com>
Closes#11736
To make use of zfs_refcount_held tunable it should be a module
parameter in open-zfs. Also, since the macros will auto-generate OS
specific tunables, removed the existing zfs_refcount_held reference
in module/os/freebsd/zfs/sysctl_os.c.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Ryan Moeller <ryan@iXsystems.com>
Reviewed-by: Allan Jude <allan@klarasystems.com>
Signed-off-by: Don Brady <don.brady@delphix.com>
Closes#11753
This will allow platforms to implement it as they see fit, in particular
in a different manner than rrm locks.
Reviewed-by: Ryan Moeller <ryan@iXsystems.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Matt Macy <mmacy@FreeBSD.org>
Signed-off-by: Mateusz Guzik <mjguzik@gmail.com>
Closes#11153
A few deadman tunables ended up in the wrong sysctl node.
Move them to vfs.zfs.deadman.*
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Alexander Motin <mav@FreeBSD.org>
Signed-off-by: Ryan Moeller <ryan@iXsystems.com>
Closes#11715
zil_replaying(zil, tx) has the side-effect of informing the ZIL that an
entry has been replayed in the (still open) tx. The ZIL uses that
information to record the replay progress in the ZIL header when that
tx's txg syncs.
ZPL log entries are not idempotent and logically dependent and thus
calling zil_replaying() is necessary for correctness.
For ZVOLs the question of correctness is more nuanced: ZVOL logs only
TX_WRITE and TX_TRUNCATE, both of which are idempotent. Logical
dependencies between two records exist only if the write or discard
request had sync semantics or if the ranges affected by the records
overlap.
Thus, at a first glance, it would be correct to restart replay from
the beginning if we crash before replay completes. But this does not
address the following scenario:
Assume one log record per LWB.
The chain on disk is
HDR -> 1:W(1, "A") -> 2:W(1, "B") -> 3:W(2, "X") -> 4:W(3, "Z")
where N:W(O, C) represents log entry number N which is a TX_WRITE of C
to offset A.
We replay 1, 2 and 3 in one txg, sync that txg, then crash.
Bit flips corrupt 2, 3, and 4.
We come up again and restart replay from the beginning because
we did not call zil_replaying() during replay.
We replay 1 again, then interpret 2's invalid checksum as the end
of the ZIL chain and call replay done.
The replayed zvol content is "AX".
If we had called zil_replaying() the HDR would have pointed to 3
and our resumed replay would not have replayed anything because
3 was corrupted, resulting in zvol content "BX".
If 3 logically depends on 2 then the replay corrupted the ZVOL_OBJ's
contents.
This patch adds the zil_replaying() calls to the replay functions.
Since the callbacks in the replay function need the zilog_t* pointer
so that they can call zil_replaying() we open the ZIL while
replaying in zvol_create_minor(). We also verify that replay has
been done when on-demand-opening the ZIL on the first modifying
bio.
Reviewed-by: Matthew Ahrens <mahrens@delphix.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Christian Schwarz <me@cschwarz.com>
Closes#11667
ZFS_READONLY represents the "DOS R/O" attribute.
When that flag is set, we should behave as if write access
were not granted by anything in the ACL. In particular:
We _must_ allow writes after opening the file r/w, then
setting the DOS R/O attribute, and writing some more.
(Similar to how you can write after fchmod(fd, 0444).)
Restore these semantics which were lost on FreeBSD when refactoring
zfs_write. To my knowledge Linux does not actually expose this flag,
but we'll need it to eventually so I've added the supporting checks.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Ryan Moeller <ryan@iXsystems.com>
Closes#11693
Even when supplied with an abd to abd_get_offset_struct(), the call
to abd_get_offset_impl() can allocate a different abd. Ensure to
call abd_fini_struct() on the abd that is not used.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Jorgen Lundman <lundman@lundman.net>
Closes#11683
When a device which is actively trimming or initializing becomes
FAULTED, and therefore no longer writable, cancel the active
TRIM or initialization. When the device is merely taken offline
with `zpool offline` then stop the operation but do not cancel it.
When the device is brought back online the operation will be
resumed if possible.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Co-authored-by: Brian Behlendorf <behlendorf1@llnl.gov>
Co-authored-by: Vipin Kumar Verma <vipin.verma@hpe.com>
Signed-off-by: Srikanth N S <srikanth.nagasubbaraoseetharaman@hpe.com>
Closes#11588
The metaslab_disable() call may block waiting for a txg sync.
Therefore it's important that vdev_rebuild_thread release the
SCL_CONFIG read lock it is holding before this call. Failure
to do so can result in the txg_sync thread getting blocked
waiting for this lock which results in a deadlock.
Reviewed-by: Mark Maybee <mark.maybee@delphix.com>
Reviewd-by: Srikanth N S <srikanth.nagasubbaraoseetharaman@hpe.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#11647
Calling vdev_free() only requires the we acquire the spa config
SCL_STATE_ALL locks, not the SCL_ALL locks. In particular, we need
need to avoid taking the SCL_CONFIG lock (included in SCL_ALL) as a
writer since this can lead to a deadlock. The txg_sync_thread() may
block in spa_txg_history_init_io() when taking the SCL_CONFIG lock
as a reading when it detects there's a pending writer.
Reviewed-by: Igor Kozhukhov <igor@dilos.org>
Reviewed-by: Mark Maybee <mark.maybee@delphix.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#11585
This change modifies the behavior of how we determine how much slop
space to use in the pool, such that now it has an upper limit. The
default upper limit is 128G, but is configurable via a tunable.
Reviewed-by: Matthew Ahrens <mahrens@delphix.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Prakash Surya <prakash.surya@delphix.com>
Closes#11023
This prevents a panic after a SLOG add/removal on the root pool followed
by a zpool scrub.
When a SLOG is removed, a hole takes its place - the vdev_ops for a hole
is vdev_hole_ops, which defines the handler functions of vdev_op_hold
and vdev_op_rele as NULL.
This bug has been reported in illumos and FreeBSD, a different trigger
in the FreeBSD report though.
Credit for this patch goes to Patrick Mooney <pmooney@pfmooney.com>
Obtained from: illumos-gate commit: c65bd18728f34725
External-issue: https://www.illumos.org/issues/12981
External-issue: https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=252396
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Rob Wing <rob.fx907@gmail.com>
Closes#11623
Making uio_impl.h the common header interface between Linux and FreeBSD
so both OS's can share a common header file. This also helps reduce code
duplication for zfs_uio_t for each OS.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Brian Atkinson <batkinson@lanl.gov>
Closes#11622
Fix regression seen in issue #11545 where checksum errors
where not being counted or showing up in a zpool event.
Reviewed-by: Matthew Ahrens <mahrens@delphix.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Don Brady <don.brady@delphix.com>
Closes#11609
