Previously, dmu_buf_will_clone() would roll back any dirty record, but
would not clean out the modified data nor reset the state before
releasing the lock. That leaves the last-written data in db_data, but
the dbuf in the wrong state.
This is eventually corrected when the dbuf state is made NOFILL, and
dbuf_noread() called (which clears out the old data), but at this point
its too late, because the lock was already dropped with that invalid
state.
Any caller acquiring the lock before the call into
dmu_buf_will_not_fill() can find what appears to be a clean, readable
buffer, and would take the wrong state from it: it should be getting the
data from the cloned block, not from earlier (unwritten) dirty data.
Even after the state was switched to NOFILL, the old data was still not
cleaned out until dbuf_noread(), which is another gap for a caller to
take the lock and read the wrong data.
This commit fixes all this by properly cleaning up the previous state
and then setting the new state before dropping the lock. The
DBUF_VERIFY() calls confirm that the dbuf is in a valid state when the
lock is down.
Sponsored-by: Klara, Inc.
Sponsored-By: OpenDrives Inc.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Pawel Jakub Dawidek <pawel@dawidek.net>
Signed-off-by: Rob Norris <rob.norris@klarasystems.com>
Closes#15566Closes#15526
Clean up code in dsl_scan_visitbp() by removing an unnecessary
alloc/free and `goto`. This has the side benefit of reducing CPU usage,
which is only really noticeable if we are not doing i/o for the leaf
blocks, like when `zfs_no_scrub_io` is set.
Reviewed-by: Alexander Motin <mav@FreeBSD.org>
Reviewed-by: Mark Maybee <mark.maybee@delphix.com>
Reviewed-by: Richard Yao <richard.yao@alumni.stonybrook.edu>
Signed-off-by: Matthew Ahrens <mahrens@delphix.com>
Closes#15549
Over its history this the dirty dnode test has been changed between
checking for a dnodes being on `os_dirty_dnodes` (`dn_dirty_link`) and
`dn_dirty_record`.
de198f2d9 Fix lseek(SEEK_DATA/SEEK_HOLE) mmap consistency
2531ce372 Revert "Report holes when there are only metadata changes"
ec4f9b8f3 Report holes when there are only metadata changes
454365bba Fix dirty check in dmu_offset_next()
66aca2473 SEEK_HOLE should not block on txg_wait_synced()
Also illumos/illumos-gate@c543ec060dillumos/illumos-gate@2bcf0248e9
It turns out both are actually required.
In the case of appending data to a newly created file, the dnode proper
is dirtied (at least to change the blocksize) and dirty records are
added. Thus, a single logical operation is represented by separate
dirty indicators, and must not be separated.
The incorrect dirty check becomes a problem when the first block of a
file is being appended to while another process is calling lseek to skip
holes. There is a small window where the dnode part is undirtied while
there are still dirty records. In this case, `lseek(fd, 0, SEEK_DATA)`
would not know that the file is dirty, and would go to
`dnode_next_offset()`. Since the object has no data blocks yet, it
returns `ESRCH`, indicating no data found, which results in `ENXIO`
being returned to `lseek()`'s caller.
Since coreutils 9.2, `cp` performs sparse copies by default, that is, it
uses `SEEK_DATA` and `SEEK_HOLE` against the source file and attempts to
replicate the holes in the target. When it hits the bug, its initial
search for data fails, and it goes on to call `fallocate()` to create a
hole over the entire destination file.
This has come up more recently as users upgrade their systems, getting
OpenZFS 2.2 as well as a newer coreutils. However, this problem has been
reproduced against 2.1, as well as on FreeBSD 13 and 14.
This change simply updates the dirty check to check both types of dirty.
If there's anything dirty at all, we immediately go to the "wait for
sync" stage, It doesn't really matter after that; both changes are on
disk, so the dirty fields should be correct.
Sponsored-by: Klara, Inc.
Sponsored-by: Wasabi Technology, Inc.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Alexander Motin <mav@FreeBSD.org>
Reviewed-by: Rich Ercolani <rincebrain@gmail.com>
Signed-off-by: Rob Norris <rob.norris@klarasystems.com>
Closes#15571Closes#15526
So that zdb (and others!) can get at the BRT on-disk structures.
Reviewed-by: Alexander Motin <mav@FreeBSD.org>
Reviewed-by: Kay Pedersen <mail@mkwg.de>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Rob Norris <robn@despairlabs.com>
Closes#15541
- Remove zsda_tx field, it is used only once.
- Remove unneeded string lengths checks, all names are terminated.
- Replace few explicit MAXNAMELEN usages with sizeof().
- Change dsname from MAXNAMELEN to ZFS_MAX_DATASET_NAME_LEN, as
expected by dsl_dataset_name(). Both are 256 bytes now, but it is
better to be safe.
This should have no functional difference.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Alexander Motin <mav@FreeBSD.org>
Sponsored by: iXsystems, Inc.
Closes#15535
Reviewed-by: Richard Yao <richard.yao@alumni.stonybrook.edu>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Don Brady <don.brady@klarasystems.com>
Closes#15536Closes#15564
- Generalize vdev_nowritecache handling by traversing through the
VDEV tree and skipping children ZIOs where not supported.
- Remove intermediate zio_null() in case of several VDEV children.
- Remove children handling from zio_ioctl(). There are no other
use cases for this code beside DKIOCFLUSHWRITECACHED, and would there
be, I doubt they would so straightforward apply to all VDEV children.
Comparing to removed previous optimization this should improve cases
of redundant ZILs/SLOGs.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: George Wilson <george.wilson@delphix.com>
Signed-off-by: Alexander Motin <mav@FreeBSD.org>
Sponsored by: iXsystems, Inc.
Closes#15515
In several places abd_zero() cleaned ABD filled at the next line.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Alexander Motin <mav@FreeBSD.org>
Sponsored by: iXsystems, Inc.
Closes#15514
Entries in the dbuf cache contribute only the size of the dbuf data to
the cache size. Attached "user" data is not counted. This can lead to
the data currently "owned" by the cache consuming more memory accounting
appears to show. In some cases (eg a metadnode data block with all child
dnode_t slots allocated), the actual size can be as much as 3x as what
the cache believes it to be.
This is arguably correct behaviour, as the cache is only tracking the
size of the dbuf data, not even the overhead of the dbuf_t. On the other
hand, in the above case of dnodes, evicting cached metadnode dbufs is
the only current way to reclaim the dnode objects, and can lead to the
situation where the dbuf cache appears to be comfortably within its
target memory window and yet is holding enormous amounts of slab memory
that cannot be reclaimed.
This commit adds a facility for a dbuf user to artificially inflate the
apparent size of the dbuf for caching purposes. This at least allows for
cache tuning to be adjusted to match something closer to the real memory
overhead.
metadnode dbufs carry a >1KiB allocation per dnode in their user data.
This informs the dbuf cache machinery of that fact, allowing it to make
better decisions when evicting dbufs.
Sponsored-by: Klara, Inc.
Sponsored-by: Wasabi Technology, Inc.
Reviewed-by: Alexander Motin <mav@FreeBSD.org>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Rob Norris <rob.norris@klarasystems.com>
Closes#15511
PR #15457 exposed weird logic in L2ARC write sizing. If it appeared
bigger than device size, instead of liming write it reset all the
system-wide tunables to their default. Aside of being excessive,
it did not actually help with the problem, still allowing infinite
loop to happen.
This patch removes the tunables reverting logic, but instead limits
L2ARC writes (or at least eviction/trim) to 1/4 of the capacity.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: George Amanakis <gamanakis@gmail.com>
Signed-off-by: Alexander Motin <mav@FreeBSD.org>
Sponsored by: iXsystems, Inc.
Closes#15519
Use goto out instead of return for early exit to make sure
snap_obj_array is freed.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Chunwei Chen <david.chen@nutanix.com>
Closes#15516
Current L2ARC write rate and headroom parameters are very conservative:
l2arc_write_max=8M and l2arc_headroom=2 (ie: a full L2ARC writes at
8 MB/s, scanning 16/32 MB of ARC tail each time; a warming L2ARC runs
at 2x these rates).
These values were selected 15+ years ago based on then-current SSDs
size, performance and endurance. Today we have multi-TB, fast and
cheap SSDs which can sustain much higher read/write rates.
For this reason, this patch increases l2arc_write_max to 32M and
l2arc_headroom to 8 (4x increase for both).
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Alexander Motin <mav@FreeBSD.org>
Signed-off-by: Gionatan Danti <g.danti@assyoma.it>
Closes#15457
This feature allows disks to be added one at a time to a RAID-Z group,
expanding its capacity incrementally. This feature is especially useful
for small pools (typically with only one RAID-Z group), where there
isn't sufficient hardware to add capacity by adding a whole new RAID-Z
group (typically doubling the number of disks).
== Initiating expansion ==
A new device (disk) can be attached to an existing RAIDZ vdev, by
running `zpool attach POOL raidzP-N NEW_DEVICE`, e.g. `zpool attach tank
raidz2-0 sda`. The new device will become part of the RAIDZ group. A
"raidz expansion" will be initiated, and the new device will contribute
additional space to the RAIDZ group once the expansion completes.
The `feature@raidz_expansion` on-disk feature flag must be `enabled` to
initiate an expansion, and it remains `active` for the life of the pool.
In other words, pools with expanded RAIDZ vdevs can not be imported by
older releases of the ZFS software.
== During expansion ==
The expansion entails reading all allocated space from existing disks in
the RAIDZ group, and rewriting it to the new disks in the RAIDZ group
(including the newly added device).
The expansion progress can be monitored with `zpool status`.
Data redundancy is maintained during (and after) the expansion. If a
disk fails while the expansion is in progress, the expansion pauses
until the health of the RAIDZ vdev is restored (e.g. by replacing the
failed disk and waiting for reconstruction to complete).
The pool remains accessible during expansion. Following a reboot or
export/import, the expansion resumes where it left off.
== After expansion ==
When the expansion completes, the additional space is available for use,
and is reflected in the `available` zfs property (as seen in `zfs list`,
`df`, etc).
Expansion does not change the number of failures that can be tolerated
without data loss (e.g. a RAIDZ2 is still a RAIDZ2 even after
expansion).
A RAIDZ vdev can be expanded multiple times.
After the expansion completes, old blocks remain with their old
data-to-parity ratio (e.g. 5-wide RAIDZ2, has 3 data to 2 parity), but
distributed among the larger set of disks. New blocks will be written
with the new data-to-parity ratio (e.g. a 5-wide RAIDZ2 which has been
expanded once to 6-wide, has 4 data to 2 parity). However, the RAIDZ
vdev's "assumed parity ratio" does not change, so slightly less space
than is expected may be reported for newly-written blocks, according to
`zfs list`, `df`, `ls -s`, and similar tools.
Sponsored-by: The FreeBSD Foundation
Sponsored-by: iXsystems, Inc.
Sponsored-by: vStack
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Mark Maybee <mark.maybee@delphix.com>
Authored-by: Matthew Ahrens <mahrens@delphix.com>
Contributions-by: Fedor Uporov <fuporov.vstack@gmail.com>
Contributions-by: Stuart Maybee <stuart.maybee@comcast.net>
Contributions-by: Thorsten Behrens <tbehrens@outlook.com>
Contributions-by: Fmstrat <nospam@nowsci.com>
Contributions-by: Don Brady <dev.fs.zfs@gmail.com>
Signed-off-by: Don Brady <dev.fs.zfs@gmail.com>
Closes#15022
When retrieving a system attribute, the size of the supplied
buffer is ignored. If the buffer is too small to hold the attribute,
sa_attr_op() will write past the end of the buffer.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Jason King <jking@racktopsystems.com>
Closes#15476
Add a dataset_kstats_rename function, and call it when renaming
a zvol on FreeBSD and Linux.
Reviewed-by: Alexander Motin <mav@FreeBSD.org>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Alan Somers <asomers@gmail.com>
Sponsored-by: Axcient
Closes#15482Closes#15486
Otherwise callbacks may trigger KMSAN violations in the dlen == 0 case.
For example, raidz_syn_pq_abd() will compare an uninitialized pointer
with itself before returning. This seems harmless, but let's maintain
good hygiene and avoid passing uninitialized variables, if only to
placate KMSAN.
Reviewed-by: Alexander Motin <mav@FreeBSD.org>
Reviewed-by: Allan Jude <allan@klarasystems.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Mark Johnston <markj@FreeBSD.org>
Closes#15491
Currently vdev_queue_class_length is responsible for checking how long
the queue length is, however, it doesn't check the length when a list
is used, rather it just returns whether it is empty or not. To fix this
I added a counter variable to vdev_queue_class to keep track of the sync
IO ops, and changed vdev_queue_class_length to reference this variable
instead.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Alexander Motin <mav@FreeBSD.org>
Signed-off-by: MigeljanImeri <ImeriMigel@gmail.com>
Closes#15478
As part of transaction group commit, dsl_pool_sync() sequentially calls
dsl_dataset_sync() for each dirty dataset, which subsequently calls
dmu_objset_sync(). dmu_objset_sync() in turn uses up to 75% of CPU
cores to run sync_dnodes_task() in taskq threads to sync the dirty
dnodes (files).
There are two problems:
1. Each ZVOL in a pool is a separate dataset/objset having a single
dnode. This means the objsets are synchronized serially, which
leads to a bottleneck of ~330K blocks written per second per pool.
2. In the case of multiple dirty dnodes/files on a dataset/objset on a
big system they will be sync'd in parallel taskq threads. However,
it is inefficient to to use 75% of CPU cores of a big system to do
that, because of (a) bottlenecks on a single write issue taskq, and
(b) allocation throttling. In addition, if not for the allocation
throttling sorting write requests by bookmarks (logical address),
writes for different files may reach space allocators interleaved,
leading to unwanted fragmentation.
The solution to both problems is to always sync no more and (if
possible) no fewer dnodes at the same time than there are allocators
the pool.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Alexander Motin <mav@FreeBSD.org>
Signed-off-by: Edmund Nadolski <edmund.nadolski@ixsystems.com>
Closes#15197
Block cloning from an encrypted dataset into an unencrypted dataset
and vice versa is not possible. The current code did allow cloning
unencrypted files into an encrypted dataset causing a panic when
these were accessed. Block cloning between encrypted and encrypted
is currently supported on the same filesystem only.
Reviewed-by: Alexander Motin <mav@FreeBSD.org>
Reviewed-by: Kay Pedersen <mail@mkwg.de>
Reviewed-by: Rob N <robn@despairlabs.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Martin Matuska <mm@FreeBSD.org>
Closes#15464Closes#15465
The change in the zvol readonly property does not update the block
device readonly entry until the first IO to the ZVOL. This patch
addresses the issue by updating the block device readonly property
from the set property IOCTL call.
Reviewed-by: Tony Hutter <hutter2@llnl.gov>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Alexander Motin <mav@FreeBSD.org>
Signed-off-by: Ameer Hamza <ahamza@ixsystems.com>
Closes#15409
Currently, zvol threading can be switched through the zvol_request_sync
module parameter system-wide. By making it a zvol property, zvol
threading can be switched per zvol.
Reviewed-by: Tony Hutter <hutter2@llnl.gov>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Alexander Motin <mav@FreeBSD.org>
Signed-off-by: Ameer Hamza <ahamza@ixsystems.com>
Closes#15409
zvol_set_volmode() and zvol_set_snapdev() share a common code path.
Merge this shared code path into zvol_set_common().
Reviewed-by: Tony Hutter <hutter2@llnl.gov>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Alexander Motin <mav@FreeBSD.org>
Signed-off-by: Ameer Hamza <ahamza@ixsystems.com>
Closes#15409
There is no sense to have separate implementations for FreeBSD and
Linux. Make Linux code shared as more functional and just register
FreeBSD-specific prune callback with arc_add_prune_callback() API.
Aside of code cleanup this should fix excessive pruning on FreeBSD:
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=274698
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Mark Johnston <markj@FreeBSD.org>
Signed-off-by: Alexander Motin <mav@FreeBSD.org>
Sponsored by: iXsystems, Inc.
Closes#15456
We should not always use PAGESIZE alignment for caches bigger than
it and SPA_MINBLOCKSIZE otherwise. Doing that caches for 5, 6, 7,
10 and 14KB rounded up to 8, 12 and 16KB respectively make no sense.
Instead specify as alignment the biggest power-of-2 divisor. This
way 2KB and 6KB caches are both aligned to 2KB, while 4KB and 8KB
are aligned to 4KB.
Reduce number of caches to half-power of 2 instead of quarter-power
of 2. This removes caches difficult for underlying allocators to
fit into page-granular slabs, such as: 2.5, 3.5, 5, 7, 10KB, etc.
Since these caches are mostly used for transient allocations like
ZIOs and small DBUF cache it does not worth being too aggressive.
Due to the above alignment issue some of those caches were not
working properly any way. 6KB cache now finally has a chance to
work right, placing 2 buffers into 3 pages, that makes sense.
Remove explicit alignment in Linux user-space case. I don't think
it should be needed any more with the above fixes.
As result on FreeBSD instead of such numbers of pages per slab:
vm.uma.zio_buf_comb_16384.keg.ppera: 4
vm.uma.zio_buf_comb_14336.keg.ppera: 4
vm.uma.zio_buf_comb_12288.keg.ppera: 3
vm.uma.zio_buf_comb_10240.keg.ppera: 3
vm.uma.zio_buf_comb_8192.keg.ppera: 2
vm.uma.zio_buf_comb_7168.keg.ppera: 2
vm.uma.zio_buf_comb_6144.keg.ppera: 2 <= Broken
vm.uma.zio_buf_comb_5120.keg.ppera: 2
vm.uma.zio_buf_comb_4096.keg.ppera: 1
vm.uma.zio_buf_comb_3584.keg.ppera: 7 <= Hard to free
vm.uma.zio_buf_comb_3072.keg.ppera: 3
vm.uma.zio_buf_comb_2560.keg.ppera: 2
vm.uma.zio_buf_comb_2048.keg.ppera: 1
vm.uma.zio_buf_comb_1536.keg.ppera: 2
vm.uma.zio_buf_comb_1024.keg.ppera: 1
vm.uma.zio_buf_comb_512.keg.ppera: 1
I am now getting such:
vm.uma.zio_buf_comb_16384.keg.ppera: 4
vm.uma.zio_buf_comb_12288.keg.ppera: 3
vm.uma.zio_buf_comb_8192.keg.ppera: 2
vm.uma.zio_buf_comb_6144.keg.ppera: 3 <= Fixed, 2 in 3 pages
vm.uma.zio_buf_comb_4096.keg.ppera: 1
vm.uma.zio_buf_comb_3072.keg.ppera: 3
vm.uma.zio_buf_comb_2048.keg.ppera: 1
vm.uma.zio_buf_comb_1536.keg.ppera: 2
vm.uma.zio_buf_comb_1024.keg.ppera: 1
vm.uma.zio_buf_comb_512.keg.ppera: 1
Reviewed-by: Allan Jude <allan@klarasystems.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Alexander Motin <mav@FreeBSD.org>
Sponsored by: iXsystems, Inc.
Closes#15452
RAIDZ parity is calculated by adding data one column at a time. It
works OK for small blocks, but for large blocks results of previous
addition may already be evicted from CPU caches to main memory, and
in addition to extra memory write require extra read to get it back.
This patch splits large parity operations into 64KB chunks, that
should in most cases fit into CPU L2 caches from the last decade.
I haven't touched more complicated cases of data reconstruction to
not over complicate the code. Those should be relatively rare.
My tests on Xeon Gold 6242R CPU with 1MB of L2 cache per core show
up to 10/20% memory traffic reduction when writing to 4-wide RAIDZ/
RAIDZ2 blocks of ~4MB and up. Older CPUs with 256KB of L2 cache
should see the effect even on smaller blocks. Wider vdevs may need
bigger blocks to be affected.
Reviewed-by: Brian Atkinson <batkinson@lanl.gov>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Alexander Motin <mav@FreeBSD.org>
Sponsored by: iXsystems, Inc.
Closes#15448
ZVOL:
- Mark all ZVOL ZIL transactions as sync. Since ZVOLs have only
one object, it makes no sense to maintain async queue and on each
commit merge it into sync. Single sync queue is just cheaper, while
it changes nothing until actual commit request arrives.
- Remove zsd_sync_cnt and the zil_async_to_sync() calls since we
are no longer switching between sync and async queues.
ZFS:
- Mark write transactions as sync based only on number of sync
opens (z_sync_cnt). We can not randomly jump between sync and
async unless we want data corruptions due to writes reordering.
- When file first opened with O_SYNC (z_sync_cnt incremented to 1)
call zil_async_to_sync() for it to preserve correct ordering between
past and future writes.
- Drop zfs_fsyncer_key logic. Looks like it was an optimization
for workloads heavily intermixing async writes with tons of fsyncs.
But first it was broken 8 years ago due to Linux tsd implementation
not allowing data storage between syscalls, and second, I doubt it
is safe to switch from async to sync so often and without calling
zil_async_to_sync().
- Rename sync argument of *_log_write() into commit, now only
signalling caller's intent to call zil_commit() soon after. It
allows WR_COPIED optimizations without extra other meanings.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: George Wilson <george.wilson@delphix.com>
Signed-off-by: Alexander Motin <mav@FreeBSD.org>
Sponsored by: iXsystems, Inc.
Closes#15366
Prefetched buffers are currently read from L2ARC if, and only if,
l2arc_noprefetch is set to non-default value of 0. This means that
a streaming read which can be served from L2ARC will instead engage
the main pool.
For example, consider what happens when a file is sequentially read:
- application requests contiguous data, engaging the prefetcher;
- ARC buffers are initially marked as prefetched but, as the calling
application consumes data, the prefetch tag is cleared;
- these "normal" buffers become eligible for L2ARC and are copied to it;
- re-reading the same file will *not* engage L2ARC even if it contains
the required buffers;
- main pool has to suffer another sequential read load, which (due to
most NCQ-enabled HDDs preferring sequential loads) can dramatically
increase latency for uncached random reads.
In other words, current behavior is to write data to L2ARC (wearing it)
without using the very same cache when reading back the same data. This
was probably useful many years ago to preserve L2ARC read bandwidth but,
with current SSD speed/size/price, it is vastly sub-optimal.
Setting l2arc_noprefetch=1, while enabling L2ARC to serve these reads,
means that even prefetched but unused buffers will be copied into L2ARC,
further increasing wear and load for potentially not-useful data.
This patch enable prefetched buffer to be read from L2ARC even when
l2arc_noprefetch=1 (default), increasing sequential read speed and
reducing load on the main pool without polluting L2ARC with not-useful
(ie: unused) prefetched data. Moreover, it clear users confusion about
L2ARC size increasing but not serving any IO when doing sequential
reads.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Alexander Motin <mav@FreeBSD.org>
Signed-off-by: Gionatan Danti <g.danti@assyoma.it>
Closes#15451
Many long-running ZFS ioctls lock the spa_namespace_lock, forcing
concurrent ioctls to sleep for the mutex. Previously, the only
option is to call mutex_enter() which sleeps uninterruptibly. This
is a usability issue for sysadmins, for example, if the admin runs
`zpool status` while a slow `zpool import` is ongoing, the admin's
shell will be locked in uninterruptible sleep for a long time.
This patch resolves this admin usability issue by introducing
mutex_enter_interruptible() which sleeps interruptibly while waiting
to acquire a lock. It is implemented for both Linux and FreeBSD.
The ZFS_IOC_POOL_CONFIGS ioctl, used by `zpool status`, is changed to
use this new macro so that the command can be interrupted if it is
issued during a concurrent `zpool import` (or other long-running
operation).
Reviewed-by: Tony Hutter <hutter2@llnl.gov>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Thomas Bertschinger <bertschinger@lanl.gov>
Closes#15360
zio_root() has no arguments for ready callback or parent ZIO. Except
one recent case in ZIL code if root ZIOs ever have a parent it is
also a root ZIO. It means we do not need READY pipeline stage for
them, which takes some time to process, but even more time to wait
for the children and be woken by them, and both for no good reason.
The most visible effect of this change is that it avoids one taskq
wakeup per ZIL block written, previously used to run zio_ready()
for lwb_root_zio and skipped now.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Alexander Motin <mav@FreeBSD.org>
Sponsored by: iXsystems, Inc.
Closes#15398
... by checking that previous block is fully written and flushed.
It allows to skip commit delays since we can give up on aggregation
in that case. This removes zil_min_commit_timeout parameter, since
for single-threaded workloads it is not needed at all, while on very
fast devices even some multi-threaded workloads may get detected as
single-threaded and still bypass the wait. To give multi-threaded
workloads more aggregation chances increase zfs_commit_timeout_pct
from 5 to 10%, as they should suffer less from additional latency.
Also single-threaded workloads detection allows in perspective better
prediction of the next block size.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Prakash Surya <prakash.surya@delphix.com>
Signed-off-by: Alexander Motin <mav@FreeBSD.org>
Sponsored by: iXsystems, Inc.
Closes#15381
Once we verified the ABDs and asserted the sizes we should never
see premature ABDs ends. Assert that and remove extra branches
from production builds.
Reviewed-by: Brian Atkinson <batkinson@lanl.gov>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Alexander Motin <mav@FreeBSD.org>
Sponsored by: iXsystems, Inc.
Closes#15428
ZFS prefetch is currently governed by the zfs_prefetch_disable
tunable. However, this is a module-wide settings - if a specific
dataset benefits from prefetch, while others have issue with it,
an optimal solution does not exists.
This commit introduce the "prefetch" tri-state property, which enable
granular control (at dataset/volume level) for prefetching.
This patch does not remove the zfs_prefetch_disable, which remains
a system-wide switch for enable/disable prefetch. However, to avoid
duplication, it would be preferable to deprecate and then remove
the module tunable.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Alexander Motin <mav@FreeBSD.org>
Reviewed-by: Ameer Hamza <ahamza@ixsystems.com>
Signed-off-by: Gionatan Danti <g.danti@assyoma.it>
Co-authored-by: Gionatan Danti <g.danti@assyoma.it>
Closes#15237Closes#15436
This reverts commit 797f55ef12 which
was causing a VERIFY failure when running the project quota tests.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#15438
spa_thread() and the "System Duty Cycle" scheduling class are from
Illumos and have not yet been adapted to Linux or FreeBSD.
HAVE_SPA_THREAD has long been explicitly undefined and used to mark
spa_thread(), but there's some related taskq code that can never be
invoked without it, which makes some already-tricky code harder to read.
HAVE_SYSDC is introduced in this commit to mark the SDC parts. SDC
requires spa_thread(), but the inverse is not true, so they are
separate.
I don't want to make the call to just remove it because I still harbour
hopes that OpenZFS could become a first-class citizen on Illumos
someday. But hopefully this will at least make the reason it exists a
bit clearer for people without long memories and/or an interest in
history.
For those that are interested in the history, the original FreeBSD port
of ZFS (before ZFS-on-Linux was adopted there) did have a spa_thread(),
but not SDC. The last version of that before it was removed can be read
here:
22df1ffd81/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/spa.c
Meanwhile, more information on the SDC scheduling class is here:
https://github.com/illumos/illumos-gate/blob/master/usr/src/uts/common/disp/sysdc.c
Sponsored-by: Klara, Inc.
Sponsored-by: Wasabi Technology, Inc.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Alexander Motin <mav@FreeBSD.org>
Signed-off-by: Rob Norris <rob.norris@klarasystems.com>
Closes#15406
In the zfs_id_over*quota functions, there is a short-circuit to skip
the zap_lookup when the quota zap does not exist. If quotas are never
used in a zpool, then the quota zap will never exist. But if
user/group/project quotas are ever used, the zap objects will be
created and will persist even if the quotas are deleted.
The quota zap_lookup in the write path can become a bottleneck for
write-heavy small I/O workloads. Before this commit, it was not
possible to remove this lookup without creating a new zpool.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Sam Atkinson <samatk@amazon.com>
Closes#14721
In my understanding ARC_BUF_SHARED() and arc_buf_is_shared() should
return identical results, except the second also asserts it deeper.
The first is much cheaper though, saving few pointer dereferences.
Replace production arc_buf_is_shared() calls with ARC_BUF_SHARED(),
and call arc_buf_is_shared() in random assertions, while making it
even more strict.
On my tests this in half reduces arc_buf_destroy_impl() time, that
noticeably reduces hash_lock congestion under heavy dbuf eviction.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: George Wilson <george.wilson@delphix.com>
Signed-off-by: Alexander Motin <mav@FreeBSD.org>
Sponsored by: iXsystems, Inc.
Closes#15397
Torn reads/writes of dp_dirty_total are unlikely: on 64-bit systems
due to register size, while on 32-bit due to memory constraints.
And even if we hit some race, the code implementing the delay takes
the lock any way.
Removal of the poll-wide lock acquisition saves ~1% of CPU time on
8-thread 8KB write workload.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Alexander Motin <mav@FreeBSD.org>
Sponsored by: iXsystems, Inc.
Closes#15390
When a vdev is to be expanded -- either via `zpool online -e` or via
the autoexpand option -- a SPA_ASYNC_CONFIG_UPDATE request is queued
to be handled via an asynchronous worker thread (spa_async_thread).
This normally happens almost immediately; but will be delayed up to
zfs_ccw_retry_interval seconds (default 5 minutes) if an attempt to
write the zpool configuration cache failed.
When FreeBSD boots ZFS-root VM images generated using `makefs -t zfs`,
the zpoolupgrade rc.d script runs `zpool upgrade`, which modifies the
pool configuration and triggers an attempt to write to the cache file.
This attempted write fails because the filesystem is still mounted
read-only at this point in the boot process, triggering a 5-minute
cooldown before SPA_ASYNC_CONFIG_UPDATE requests will be handled by
the asynchronous worker thread.
When expanding a vdev, reset the "when did a configuration cache
write last fail" value so that the SPA_ASYNC_CONFIG_UPDATE request
will be handled promptly. A cleaner but more intrusive option would
be to use separate SPA_ASYNC_ flags for "configuration changed" and
"try writing the configuration cache again", but with FreeBSD 14.0
coming very soon I'd prefer to leave such refactoring for a later
date.
Reviewed-by: Alexander Motin <mav@FreeBSD.org>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Colin Percival <cperciva@FreeBSD.org>
Closes#15405
For synchronous write workloads with large IO sizes, a pool configured
with a slog performs worse than one with an embedded zil:
sequential_writes 1m sync ios, 16 threads
Write IOPS: 1292 438 -66.10%
Write Bandwidth: 1323570 448910 -66.08%
Write Latency: 12128400 36330970 3.0x
sequential_writes 1m sync ios, 32 threads
Write IOPS: 1293 430 -66.74%
Write Bandwidth: 1324184 441188 -66.68%
Write Latency: 24486278 74028536 3.0x
The reason is the `zil_slog_bulk` variable. In `zil_lwb_write_open`,
if a zil block is greater than 768K, the priority of the write is
downgraded from sync to async. Increasing the value allows greater
throughput. To select a value for this PR, I ran an fio workload with
the following values for `zil_slog_bulk`:
zil_slog_bulk KiB/s
1048576 422132
2097152 478935
4194304 533645
8388608 623031
12582912 827158
16777216 1038359
25165824 1142210
33554432 1211472
50331648 1292847
67108864 1308506
100663296 1306821
134217728 1304998
At 64M, the results with a slog are now improved to parity with an
embedded zil:
sequential_writes 1m sync ios, 16 threads
Write IOPS: 438 1288 2.9x
Write Bandwidth: 448910 1319062 2.9x
Write Latency: 36330970 12163408 -66.52%
sequential_writes 1m sync ios, 32 threads
Write IOPS: 430 1290 3.0x
Write Bandwidth: 441188 1321693 3.0x
Write Latency: 74028536 24519698 -66.88%
None of the other tests in the performance suite (run with a zil or
slog) had a significant change, including the random_write_zil tests,
which use multiple datasets.
Reviewed-by: Alexander Motin <mav@FreeBSD.org>
Reviewed-by: Tony Nguyen <tony.nguyen@delphix.com>
Signed-off-by: John Wren Kennedy <john.kennedy@delphix.com>
Closes#14378
When doing a manual TRIM on a zpool, the metaslab being TRIMmed is
potentially re-enabled before all queued TRIM zios for that metaslab
have completed. Since TRIM zios have the lowest priority, it is
possible to get into a situation where allocations occur from the
just re-enabled metaslab and cut ahead of queued TRIMs to the same
metaslab. If the ranges overlap, this will cause corruption.
We were able to trigger this pretty consistently with a small single
top-level vdev zpool (i.e. small number of metaslabs) with heavy
parallel write activity while performing a manual TRIM against a
somewhat 'slow' device (so TRIMs took a bit of time to complete).
With the patch, we've not been able to recreate it since. It was on
illumos, but inspection of the OpenZFS trim code looks like the
relevant pieces are largely unchanged and so it appears it would be
vulnerable to the same issue.
Reviewed-by: Igor Kozhukhov <igor@dilos.org>
Reviewed-by: Alexander Motin <mav@FreeBSD.org>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Jason King <jking@racktopsystems.com>
Illumos-issue: https://www.illumos.org/issues/15939Closes#15395
dmu_tx_check_ioerr() pre-reads blocks that are going to be dirtied
as part of transaction to both prefetch them and check for errors.
But it makes no sense to do it for holes, since there are no disk
reads to prefetch and there can be no errors. On the other side
those blocks are anonymous, and they are freed immediately by the
dbuf_rele() without even being put into dbuf cache, so we just
burn CPU time on decompression and overheads and get absolutely
no result at the end.
Use of dbuf_hold_impl() with fail_sparse parameter allows to skip
the extra work, and on my tests with sequential 8KB writes to empty
ZVOL with 32KB blocks shows throughput increase from 1.7 to 2GB/s.
Reviewed-by: Brian Atkinson <batkinson@lanl.gov>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Alexander Motin <mav@FreeBSD.org>
Sponsored by: iXsystems, Inc.
Closes#15371
Benchmarks show that at certain write sizes range lock/unlock take
not so much time as extra memory copy. The exact threshold is not
obvious due to other overheads, but it is definitely lower than
~63KB used before. Make it configurable, defaulting at 7.5KB,
that is 8KB of nearest malloc() size minus itx and lr structs.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Alexander Motin <mav@FreeBSD.org>
Sponsored by: iXsystems, Inc.
Closes#15353
Previously, if a cachefile is passed to zpool import, the cached config
is mostly offered as-is to ZFS_IOC_POOL_TRYIMPORT->spa_tryimport(), and
the results are taken as the canonical pool config and handed back to
ZFS_IOC_POOL_IMPORT.
In the course of its operation, spa_load() will inspect the pool and
build a new config from what it finds on disk. However, it then
regenerates a new config ready to import, and so rightly sets the hostid
and hostname for the local host in the config it returns.
Because of this, the "require force" checks always decide the pool is
exported and last touched by the local host, even if this is not true,
which is possible in a HA environment when MMP is not enabled. The pool
may be imported on another head, but the import checks still pass here,
so the pool ends up imported on both.
(This doesn't happen when a cachefile isn't used, because the pool
config is discovered in userspace in zpool_find_import(), and that does
find the on-disk hostid and hostname correctly).
Since the systemd zfs-import-cache.service unit uses cachefile imports,
this can lead to a system returning after a crash with a "valid"
cachefile on disk and automatically, quietly, importing a pool that has
already been taken up by a secondary head.
This commit causes the on-disk hostid and hostname to be included in the
ZPOOL_CONFIG_LOAD_INFO item in the returned config, and then changes the
"force" checks for zpool import to use them if present.
This method should give no change in behaviour for old userspace on new
kernels (they won't know to look for the new config items) and for new
userspace on old kernels (the won't find the new config items).
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Rob Norris <rob.norris@klarasystems.com>
Sponsored-by: Klara, Inc.
Sponsored-by: Wasabi Technology, Inc.
Closes#15290
Before this change ZFS created threads for 50% of CPUs for each top-
level vdev. Plus it created the same number of threads for embedded
log groups (that have only one metaslab and don't need any preload).
As result, on system with 80 CPUs and pool of 60 vdevs this resulted
in 4800 metaslab preload threads, that is absolutely insane.
This patch changes the preload threads to 50% of CPUs in one taskq
per pool, so on the mentioned system it will be only 40 threads.
Among other things this fixes zdb on the mentioned system and pool
on FreeBSD, that failed to create so many threads in one process.
Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Alexander Motin <mav@FreeBSD.org>
Sponsored by: iXsystems, Inc.
Closes#15319
To reduce memory usage ZFS crypto allocated bigger by 56 bytes ARC
headers only when specific block was encrypted on disk. It was a
nice optimization, except in some cases the code reallocated them
on fly, that invalidated header pointers from the buffers. Since
the buffers use different locking, it created number of races, that
were originally covered (at least partially) by b_evict_lock, used
also to protection evictions. But it has gone as part of #14340.
As result, as was found in #15293, arc_hdr_realloc_crypt() ended
up unprotected and causing use-after-free.
Instead of introducing some even more elaborate locking, this patch
just drops the difference between normal and protected headers. It
cost us additional 56 bytes per header, but with couple patches
saving 24 bytes, the net growth is only 32 bytes with total header
size of 232 bytes on FreeBSD, that IMHO is acceptable price for
simplicity. Additional locking would also end up consuming space,
time or both.
Reviewe-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Alexander Motin <mav@FreeBSD.org>
Sponsored by: iXsystems, Inc.
Closes#15293Closes#15347
In most cases we do not care about exact number of buffers linked
to the header, we just need to know if it is zero, non-zero or one.
That can easily be checked just looking on b_buf pointer or in some
cases derefencing it.
b_ebufcnt is read only once, and in that case we already traverse
the list as part of arc_buf_remove(), so second traverse should not
be expensive.
This reduces L1 ARC header size by 8 bytes and full crypto header by
16 bytes, down to 176 and 232 bytes on FreeBSD respectively.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Alexander Motin <mav@FreeBSD.org>
Sponsored by: iXsystems, Inc.
Closes#15350
Earlier as part of #14123 I've removed one use of b_cv. This patch
reuses the same approach to remove the other one from much more
rare code path.
This saves 16 bytes of L1 ARC header on FreeBSD (reducing it from
200 to 184 bytes) and seems even more on Linux.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Alexander Motin <mav@FreeBSD.org>
Sponsored by: iXsystems, Inc.
Closes#15340
Commit 8af1104f does not actually store the ashift of cache devices in
their label. However, in order to facilitate reporting the ashift
through zdb, we enable this in the present commit. We also document
how the retrieval of the ashift is done.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: George Amanakis <gamanakis@gmail.com>
Closes#15331
If we are copying only one block and it is smaller than recordsize
property, do not allow destination to grow beyond one block if it
is not there yet. Otherwise the destination will get stuck with
that block size forever, that can be as small as 512 bytes, no
matter how big the destination grow later.
Reviewed-by: Kay Pedersen <mail@mkwg.de>
Reviewed-by: Rob Norris <rob.norris@klarasystems.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Alexander Motin <mav@FreeBSD.org>
Sponsored by: iXsystems, Inc.
Closes#15321
Vendor testing shows we should be able to get a little more
performance if we further relax the hard limit which we're hitting.
Authored-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Alexander Motin <mav@FreeBSD.org>
Signed-off-by: Tony Hutter <hutter2@llnl.gov>
Closes#15324
