Mixing ZIL and normal allocations has several problems:
1. The ZIL allocations are allocated, written to disk, and then a few
seconds later freed. This leaves behind holes (free segments) where the
ZIL blocks used to be, which increases fragmentation, which negatively
impacts performance.
2. When under moderate load, ZIL allocations are of 128KB. If the pool
is fairly fragmented, there may not be many free chunks of that size.
This causes ZFS to load more metaslabs to locate free segments of 128KB
or more. The loading happens synchronously (from zil_commit()), and can
take around a second even if the metaslab's spacemap is cached in the
ARC. All concurrent synchronous operations on this filesystem must wait
while the metaslab is loading. This can cause a significant performance
impact.
3. If the pool is very fragmented, there may be zero free chunks of
128KB or more. In this case, the ZIL falls back to txg_wait_synced(),
which has an enormous performance impact.
These problems can be eliminated by using a dedicated log device
("slog"), even one with the same performance characteristics as the
normal devices.
This change sets aside one metaslab from each top-level vdev that is
preferentially used for ZIL allocations (vdev_log_mg,
spa_embedded_log_class). From an allocation perspective, this is
similar to having a dedicated log device, and it eliminates the
above-mentioned performance problems.
Log (ZIL) blocks can be allocated from the following locations. Each
one is tried in order until the allocation succeeds:
1. dedicated log vdevs, aka "slog" (spa_log_class)
2. embedded slog metaslabs (spa_embedded_log_class)
3. other metaslabs in normal vdevs (spa_normal_class)
The space required for the embedded slog metaslabs is usually between
0.5% and 1.0% of the pool, and comes out of the existing 3.2% of "slop"
space that is not available for user data.
On an all-ssd system with 4TB storage, 87% fragmentation, 60% capacity,
and recordsize=8k, testing shows a ~50% performance increase on random
8k sync writes. On even more fragmented systems (which hit problem #3
above and call txg_wait_synced()), the performance improvement can be
arbitrarily large (>100x).
Reviewed-by: Serapheim Dimitropoulos <serapheim@delphix.com>
Reviewed-by: George Wilson <gwilson@delphix.com>
Reviewed-by: Don Brady <don.brady@delphix.com>
Reviewed-by: Mark Maybee <mark.maybee@delphix.com>
Signed-off-by: Matthew Ahrens <mahrens@delphix.com>
Closes#11389
In FreeBSD the struct uio was just a typedef to uio_t. In order to
extend this struct, outside of the definition for the struct uio, the
struct uio has been embedded inside of a uio_t struct.
Also renamed all the uio_* interfaces to be zfs_uio_* to make it clear
this is a ZFS interface.
Reviewed-by: Ryan Moeller <ryan@iXsystems.com>
Reviewed-by: Jorgen Lundman <lundman@lundman.net>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Brian Atkinson <batkinson@lanl.gov>
Closes#11438
The `abd_get_offset_*()` routines create an abd_t that references
another abd_t, and doesn't allocate any pages/buffers of its own. In
some workloads, these routines may be called frequently, to create many
abd_t's representing small pieces of a single large abd_t. In
particular, the upcoming RAIDZ Expansion project makes heavy use of
these routines.
This commit adds the ability for the caller to allocate and provide the
abd_t struct to a variant of `abd_get_offset_*()`. This eliminates the
cost of allocating the abd_t and performing the accounting associated
with it (`abdstat_struct_size`). The RAIDZ/DRAID code uses this for
the `rc_abd`, which references the zio's abd. The upcoming RAIDZ
Expansion project will leverage this infrastructure to increase
performance of reads post-expansion by around 50%.
Additionally, some of the interfaces around creating and destroying
abd_t's are cleaned up. Most significantly, the distinction between
`abd_put()` and `abd_free()` is eliminated; all types of abd_t's are
now disposed of with `abd_free()`.
Reviewed-by: Brian Atkinson <batkinson@lanl.gov>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Matthew Ahrens <mahrens@delphix.com>
Issue #8853Closes#11439
Each zfs ioctl that changes on-disk state (e.g. set property, create
snapshot, destroy filesystem) is recorded in the zpool history, and is
printed by `zpool history -i`.
For performance diagnostic purposes, it would be useful to know how long
each of these ioctls took to run. This commit adds that functionality,
with a new `ZPOOL_HIST_ELAPSED_NS` member of the history nvlist.
Additionally, the time recorded in this history log is currently the
time that the history record is written to disk. But in many cases (CLI
args logging and ioctl logging), this happens asynchronously,
potentially many seconds after the operation completed. This commit
changes the timestamp to reflect when the history event was created,
rather than when it was written to disk.
Reviewed-by: Mark Maybee <mmaybee@cray.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Matthew Ahrens <mahrens@delphix.com>
Closes#11440
If the system is very low on memory (specifically,
`arc_free_memory() < arc_sys_free/2`, i.e. less than 1/16th of RAM
free), `arc_evict_state_impl()` will defer wakups. In this case, the
arc_evict_waiter_t's remain on the list, even though `arc_evict_count`
has been incremented past their `aew_count`.
The problem is that `arc_wait_for_eviction()` assumes that if there are
waiters on the list, the count they are waiting for has not yet been
reached. However, the deferred wakeups may violate this, causing
`ASSERT(last->aew_count > arc_evict_count)` to fail.
This commit resolves the issue by having new waiters use the greater of
`arc_evict_count` and the last `aew_count`.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: George Wilson <gwilson@delphix.com>
Reviewed-by: George Amanakis <gamanakis@gmail.com>
Signed-off-by: Matthew Ahrens <mahrens@delphix.com>
Closes#11285Closes#11397
Build error on illumos with gcc 10 did reveal:
In function 'dmu_objset_refresh_ownership':
../../common/fs/zfs/dmu_objset.c:857:25: error: implicit conversion
from 'boolean_t' to 'ds_hold_flags_t' {aka 'enum ds_hold_flags'}
[-Werror=enum-conversion]
857 | dsl_dataset_disown(ds, decrypt, tag);
| ^~~~~~~
cc1: all warnings being treated as errors
libzfs_input_check.c: In function 'zfs_ioc_input_tests':
libzfs_input_check.c:754:28: error: implicit conversion from
'enum dmu_objset_type' to 'enum lzc_dataset_type'
[-Werror=enum-conversion]
754 | err = lzc_create(dataset, DMU_OST_ZFS, NULL, NULL, 0);
| ^~~~~~~~~~~
cc1: all warnings being treated as errors
The same issue is present in openzfs, and also the same issue about
ds_hold_flags_t, which currently defines exactly one valid value.
Reviewed-by: Igor Kozhukhov <igor@dilos.org>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Toomas Soome <tsoome@me.com>
Closes#11406
Individual transactions may not be larger than DMU_MAX_ACCESS.
This is enforced by the assertions in dmu_tx_hold_write() and
dmu_tx_hold_write_by_dnode(). There's an additional check in
dmu_tx_count_write() however it has no effect and only sets a
local err variable. We could enable this check, however since
it's already enforced by ASSERTs elsewhere I opted to remove it
instead.
Reviewed-by: Matthew Ahrens <mahrens@delphix.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#3731Closes#11384
After porting the fix for https://github.com/openzfs/zfs/issues/5295
over to illumos, we started hitting an assertion failure when running
the testsuite:
assertion failed: rc->rc_count == number, file: .../refcount.c
and the unexpected hold has this stack:
dsl_dataset_long_hold+0x59 dmu_objset_upgrade+0x73
dmu_objset_id_quota_upgrade+0x15 dmu_objset_own+0x14f
The simplest reproducer for this in illumos is
zpool create -f -O version=1 testpool c3t0d0; zpool destroy testpool
which is run as part of the zpool_create_tempname test, but I can't get
this to trigger on FreeBSD. This appears to be because of the call to
txg_wait_synced() in dmu_objset_upgrade_stop() (which was missing in
illumos), slows down dmu_objset_disown() enough to avoid the condition.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Andy Fiddaman <andy@omnios.org>
Closes#11368
Based on a conversation with Matt on the OpenZFS Slack.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Matthew Ahrens <mahrens@delphix.com>
Signed-off-by: Christian Schwarz <me@cschwarz.com>
Closes#11370
As of the 5.10 kernel the generic splice compatibility code has been
removed. All filesystems are now responsible for registering a
->splice_read and ->splice_write callback to support this operation.
The good news is the VFS provided generic_file_splice_read() and
iter_file_splice_write() callbacks can be used provided the ->iter_read
and ->iter_write callback support pipes. However, this is currently
not the case and only iovecs and bvecs (not pipes) are ever attached
to the uio structure.
This commit changes that by allowing full iov_iter structures to be
attached to uios. Ever since the 4.9 kernel the iov_iter structure
has supported iovecs, kvecs, bvevs, and pipes so it's desirable to
pass the entire thing when possible. In conjunction with this the
uio helper functions (i.e uiomove(), uiocopy(), etc) have been
updated to understand the new UIO_ITER type.
Note that using the kernel provided uio_iter interfaces allowed the
existing Linux specific uio handling code to be simplified. When
there's no longer a need to support kernel's older than 4.9, then
it will be possible to remove the iovec and bvec members from the
uio structure and always use a uio_iter. Until then we need to
maintain all of the existing types for older kernels.
Some additional refactoring and cleanup was included in this change:
- Added checks to configure to detect available iov_iter interfaces.
Some are available all the way back to the 3.10 kernel and are used
when available. In particular, uio_prefaultpages() now always uses
iov_iter_fault_in_readable() which is available for all supported
kernels.
- The unused UIO_USERISPACE type has been removed. It is no longer
needed now that the uio_seg enum is platform specific.
- Moved zfs_uio.c from the zcommon.ko module to the Linux specific
platform code for the zfs.ko module. This gets it out of libzfs
where it was never needed and keeps this Linux specific code out
of the common sources.
- Removed unnecessary O_APPEND handling from zfs_iter_write(), this
is redundant and O_APPEND is already handled in zfs_write();
Reviewed-by: Colin Ian King <colin.king@canonical.com>
Reviewed-by: Tony Hutter <hutter2@llnl.gov>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#11351
The space in special devices is not included in spa_dspace (or
dsl_pool_adjustedsize(), or the zfs `available` property). Therefore
there is always at least as much free space in the normal class, as
there is allocated in the special class(es). And therefore, there is
always enough free space to remove a special device.
However, the checks for free space when removing special devices did not
take this into account. This commit corrects that.
Reviewed-by: Ryan Moeller <ryan@iXsystems.com>
Reviewed-by: Don Brady <don.brady@delphix.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Matthew Ahrens <mahrens@delphix.com>
Closes#11329
After e357046 it should not be necessary to periodically update ARC
kstats and tunables. Tunable updates are applied when modified, and
kstats are updated on demand.
Update kstats in `arc_evict_cb_check()` for `ZFS_DEBUG` builds only.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Ryan Moeller <ryan@iXsystems.com>
Closes#11237
On a system with very high fragmentation, we may need to do lots of gang
allocations (e.g. most indirect block allocations (~50KB) may need to
gang). Before failing a "normal" allocation and resorting to ganging, we
try every metaslab. This has the impact of loading every metaslab (not
a huge deal since we now typically keep all metaslabs loaded), and also
iterating over every metaslab for every failing allocation. If there are
many metaslabs (more than the typical ~200, e.g. due to vdev expansion
or very large vdevs), the CPU cost of this iteration can be very
impactful. This iteration is done with the mg_lock held, creating long
hold times and high lock contention for concurrent allocations,
ultimately causing long txg sync times and poor application performance.
To address this, this commit changes the behavior of "normal" (not
try_hard, not ZIL) allocations. These will now only examine the 100
best metaslabs (as determined by their ms_weight). If none of these
have a large enough free segment, then the allocation will fail and
we'll fall back on ganging.
To accomplish this, we will now (normally) gang before doing a
`try_hard` allocation. Non-try_hard allocations will only examine the
100 best metaslabs of each vdev. In summary, we will first try normal
allocation. If that fails then we will do a gang allocation. If that
fails then we will do a "try hard" gang allocation. If that fails then
we will have a multi-layer gang block.
Reviewed-by: Paul Dagnelie <pcd@delphix.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Matthew Ahrens <mahrens@delphix.com>
Closes#11327
Metaslab rotor and aliquot are used to distribute workload between
vdevs while keeping some locality for logically adjacent blocks. Once
multiple allocators were introduced to separate allocation of different
objects it does not make much sense for different allocators to write
into different metaslabs of the same metaslab group (vdev) same time,
competing for its resources. This change makes each allocator choose
metaslab group independently, colliding with others only sporadically.
Test including simultaneous write into 4 files with recordsize of 4KB
on a striped pool of 30 disks on a system with 40 logical cores show
reduction of vdev queue lock contention from 54 to 27% due to better
load distribution. Unfortunately it won't help much ZVOLs yet since
only one dataset/ZVOL is synced at a time, and so for the most part
only one allocator is used, but it may improve later.
While there, to reduce the number of pointer dereferences change
per-allocator storage for metaslab classes and groups from several
separate malloc()'s to variable length arrays at the ends of the
original class and group structures.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Matthew Ahrens <mahrens@delphix.com>
Signed-off-by: Alexander Motin <mav@FreeBSD.org>
Closes#11288
The last change caused the read completion callback to not be called
if the IO was still in progress. This change restores allocation
of the arc buf callback, but in the callback path checks the new
acb_nobuf field to know to skip buffer allocation.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Matt Macy <mmacy@FreeBSD.org>
Closes#11324
When removing and subsequently reattaching a vdev, CKSUM errors may
occur as vdev_indirect_read_all() reads from all children of a mirror
in case of a resilver.
Fix this by checking whether a child is missing the data and setting a
flag (ic_error) which is then checked in vdev_indirect_repair() and
suppresses incrementing the checksum counter.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: George Amanakis <gamanakis@gmail.com>
Closes#11277
The performance of `zfs receive` can be bottlenecked on the CPU consumed
by the `receive_writer` thread, especially when receiving streams with
small compressed block sizes. Much of the CPU is spent creating and
destroying dbuf's and arc buf's, one for each `WRITE` record in the send
stream.
This commit introduces the concept of "lightweight writes", which allows
`zfs receive` to write to the DMU by providing an ABD, and instantiating
only a new type of `dbuf_dirty_record_t`. The dbuf and arc buf for this
"dirty leaf block" are not instantiated.
Because there is no dbuf with the dirty data, this mechanism doesn't
support reading from "lightweight-dirty" blocks (they would see the
on-disk state rather than the dirty data). Since the dedup-receive code
has been removed, `zfs receive` is write-only, so this works fine.
Because there are no arc bufs for the received data, the received data
is no longer cached in the ARC.
Testing a receive of a stream with average compressed block size of 4KB,
this commit improves performance by 50%, while also reducing CPU usage
by 50% of a CPU. On a per-block basis, CPU consumed by receive_writer()
and dbuf_evict() is now 1/7th (14%) of what it was.
Baseline: 450MB/s, CPU in receive_writer() 40% + dbuf_evict() 35%
New: 670MB/s, CPU in receive_writer() 17% + dbuf_evict() 0%
The code is also restructured in a few ways:
Added a `dr_dnode` field to the dbuf_dirty_record_t. This simplifies
some existing code that no longer needs `DB_DNODE_ENTER()` and related
routines. The new field is needed by the lightweight-type dirty record.
To ensure that the `dr_dnode` field remains valid until the dirty record
is freed, we have to ensure that the `dnode_move()` doesn't relocate the
dnode_t. To do this we keep a hold on the dnode until it's zio's have
completed. This is already done by the user-accounting code
(`userquota_updates_task()`), this commit extends that so that it always
keeps the dnode hold until zio completion (see `dnode_rele_task()`).
`dn_dirty_txg` was previously zeroed when the dnode was synced. This
was not necessary, since its meaning can be "when was this dnode last
dirtied". This change simplifies the new `dnode_rele_task()` code.
Removed some dead code related to `DRR_WRITE_BYREF` (dedup receive).
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Paul Dagnelie <pcd@delphix.com>
Reviewed-by: George Wilson <gwilson@delphix.com>
Signed-off-by: Matthew Ahrens <mahrens@delphix.com>
Closes#11105
In the redaction list traversal code, there is a bug in the binary search
logic when looking for the resume point. Maxbufid can be decremented to -1,
causing us to read the last possible block of the object instead of the one we
wanted. This can cause incorrect resume behavior, or possibly even a hang in
some cases. In addition, when examining non-last blocks, we can treat the
block as being the same size as the last block, causing us to miss entries in
the redaction list when determining where to resume. Finally, we were ignoring
the case where the resume point was found in the buffer being searched, and
resuming from minbufid. All these issues have been corrected, and the code has
been significantly simplified to make future issues less likely.
Reviewed-by: Serapheim Dimitropoulos <serapheim@delphix.com>
Reviewed-by: Matthew Ahrens <mahrens@delphix.com>
Signed-off-by: Paul Dagnelie <pcd@delphix.com>
Closes#11297
ZFS currently doesn't react to hotplugging cpu or memory into the
system in any way. This patch changes that by adding logic to the ARC
that allows the system to take advantage of new memory that is added
for caching purposes. It also adds logic to the taskq infrastructure
to support dynamically expanding the number of threads allocated to a
taskq.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Co-authored-by: Matthew Ahrens <matthew.ahrens@delphix.com>
Co-authored-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Paul Dagnelie <pcd@delphix.com>
Closes#11212
Add ARC_FLAG_NO_BUF to indicate that a buffer need not be
instantiated. This fixes a ~20% performance regression on
cached reads due to zfetch changes.
Reviewed-by: Tony Nguyen <tony.nguyen@delphix.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Matthew Ahrens <mahrens@delphix.com>
Signed-off-by: Matt Macy <mmacy@FreeBSD.org>
Closes#11220Closes#11232
The fnvlist_lookup_boolean_value() function should not be used
to check the force argument since it's optional. It may not be
provided or may have been created with the wrong flags.
Reviewed-by: Ryan Moeller <ryan@iXsystems.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#11281Closes#11284
During module load time all of the available fetcher4 and raidz
implementations are benchmarked for a fixed amount of time to
determine the fastest available. Manual testing has shown that this
time can be significantly reduced with negligible effect on the final
results.
This commit changes the benchmark time to 1ms which can reduce the
module load time by over a second on x86_64. On an x86_64 system
with sse3, ssse3, and avx2 instructions the benchmark times are:
Fletcher4 603ms -> 15ms
RAIDZ 1,322ms -> 64ms
Reviewed-by: Matthew Macy <mmacy@freebsd.org>
Reviewed-by: George Melikov <mail@gmelikov.ru>
Reviewed-by: Ryan Moeller <ryan@iXsystems.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#11282
Since 8c4fb36a24 (PR #7795) spa_has_pending_synctask() started to
take two more locks per write inside txg_all_lists_empty(). I am
surprised those pool-wide locks are not contended, but still their
operations are visible in CPU profiles under contended vdev lock.
This commit slightly changes vdev_queue_max_async_writes() flow to
not call the function if we are going to return max_active any way
due to high amount of dirty data. It allows to save some CPU time
exactly when the pool is busy.
Reviewed-by: Ryan Moeller <ryan@iXsystems.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-By: Tom Caputi <caputit1@tcnj.edu>
Signed-off-by: Alexander Motin <mav@FreeBSD.org>
Closes#11280
When sending raw encrypted datasets the user space accounting is present
when it's not expected to be. This leads to the subsequent mount failure
due a checksum error when verifying the local mac.
Fix this by clearing the OBJSET_FLAG_USERACCOUNTING_COMPLETE and reset
the local mac. This allows the user accounting to be correctly updated
on first mount using the normal upgrade process.
Reviewed-By: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-By: Tom Caputi <caputit1@tcnj.edu>
Signed-off-by: George Amanakis <gamanakis@gmail.com>
Closes#10523Closes#11221
It was found that setting min_active tunables for non-interactive I/Os
makes them stuck. It is caused by zfs_vdev_nia_delay, that can never
be reached if we never issue any I/Os due to min_active set to zero.
Fix this by issuing at least one non-interactive I/O at a time when
there are no interactive I/Os. When there are interactive I/Os, zero
min_active allows to completely block any non-interactive I/O. It may
min_active starvation in some scenarios, but who we are to deny foot
shooting?
Reviewed-by: George Melikov <mail@gmelikov.ru>
Reviewed-by: Ryan Moeller <ryan@iXsystems.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Alexander Motin <mav@FreeBSD.org>
Closes#11261
This is needed for zfsd to autoreplace vdevs.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Alexander Motin <mav@FreeBSD.org>
Signed-off-by: Ryan Moeller <ryan@iXsystems.com>
Closes#11260
In function dmu_buf_hold_array_by_dnode, the usage of zio is only for
the reading operation. Only create the zio and wait it in the reading
scenario as a performance optimization.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Finix Yan <yancw@info2soft.com>
Closes#11251Closes#11256
Commit 85703f6 added a new ASSERT to zfs_write() as part of the
cleanup which isn't correct in the case where multiple processes
are concurrently extending a file. The `zp->z_size` is updated
atomically while holding a range lock on only a portion of the
file. Therefore, it's possible for the file size to increase
after a same check is performed earlier in the loop causing this
ASSERT to fail. The code itself handles this case correctly so
only the invalid ASSERT needs to be removed.
Reviewed-by: Brian Atkinson <batkinson@lanl.gov>
Reviewed-by: Ryan Moeller <ryan@iXsystems.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#11235
Investigating influence of scrub (especially sequential) on random read
latency I've noticed that on some HDDs single 4KB read may take up to 4
seconds! Deeper investigation shown that many HDDs heavily prioritize
sequential reads even when those are submitted with queue depth of 1.
This patch addresses the latency from two sides:
- by using _min_active queue depths for non-interactive requests while
the interactive request(s) are active and few requests after;
- by throttling it further if no interactive requests has completed
while configured amount of non-interactive did.
While there, I've also modified vdev_queue_class_to_issue() to give
more chances to schedule at least _min_active requests to the lowest
priorities. It should reduce starvation if several non-interactive
processes are running same time with some interactive and I think should
make possible setting of zfs_vdev_max_active to as low as 1.
I've benchmarked this change with 4KB random reads from ZVOL with 16KB
block size on newly written non-fragmented pool. On fragmented pool I
also saw improvements, but not so dramatic. Below are log2 histograms
of the random read latency in milliseconds for different devices:
4 2x mirror vdevs of SATA HDD WDC WD20EFRX-68EUZN0 before:
0, 0, 2, 1, 12, 21, 19, 18, 10, 15, 17, 21
after:
0, 0, 0, 24, 101, 195, 419, 250, 47, 4, 0, 0
, that means maximum latency reduction from 2s to 500ms.
4 2x mirror vdevs of SATA HDD WDC WD80EFZX-68UW8N0 before:
0, 0, 2, 31, 38, 28, 18, 12, 17, 20, 24, 10, 3
after:
0, 0, 55, 247, 455, 470, 412, 181, 36, 0, 0, 0, 0
, i.e. from 4s to 250ms.
1 SAS HDD SEAGATE ST14000NM0048 before:
0, 0, 29, 70, 107, 45, 27, 1, 0, 0, 1, 4, 19
after:
1, 29, 681, 1261, 676, 1633, 67, 1, 0, 0, 0, 0, 0
, i.e. from 4s to 125ms.
1 SAS SSD SEAGATE XS3840TE70014 before (microseconds):
0, 0, 0, 0, 0, 0, 0, 0, 70, 18343, 82548, 618
after:
0, 0, 0, 0, 0, 0, 0, 0, 283, 92351, 34844, 90
I've also measured scrub time during the test and on idle pools. On
idle fragmented pool I've measured scrub getting few percent faster
due to use of QD3 instead of QD2 before. On idle non-fragmented pool
I've measured no difference. On busy non-fragmented pool I've measured
scrub time increase about 1.5-1.7x, while IOPS increase reached 5-9x.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Matthew Ahrens <mahrens@delphix.com>
Reviewed-by: Ryan Moeller <ryan@iXsystems.com>
Signed-off-by: Alexander Motin <mav@FreeBSD.org>
Sponsored-By: iXsystems, Inc.
Closes#11166
Commit a1d477c2 accidentally disabled DTL updates for the zil_claim()
case described at the end of vdev_stat_update() by unconditionally
disabling all DTL updates when loading. This was done to avoid
a deadlock on the vd_dtl_lock when loading the DTLs from disk.
vdev_dtl_contains <--- Takes vd->vd_dtl_lock
vdev_mirror_child_missing
vdev_mirror_io_start
zio_vdev_io_start
__zio_execute
arc_read
dbuf_issue_final_prefetch
dbuf_prefetch_impl
dbuf_prefetch
dmu_prefetch
space_map_iterate
space_map_load_length
space_map_load
vdev_dtl_load <--- Takes vd->vd_dtl_lock
vdev_load
spa_ld_load_vdev_metadata
spa_tryimport
The missing DTL updates can be restored by moving the space_map_load()
call outside the vd_dtl_lock. A private range tree is populated by
reading the space map and then merged in to the DTL_MISSING tree
under the lock.
Furthermore, the SPA_LOAD_NONE check in vdev_dtl_contains() leads to an
additional problem. Any resilvering which occurs before SPA_LOAD_NONE
is set will incorrectly determine that there's nothing to repair. This
can result in full redundancy not being restored for some blocks.
Reviewed-by: Matt Ahrens <matt@delphix.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#11218
Is this block when abuf != NULL ever reached? Yes, it is.
Add asserts and comments to prove that when we get here, we have a full
block write at an aligned offset extending past EOF.
Simplify by removing the check that tx_bytes == max_blksz, since we can
assert that it is always true.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Ryan Moeller <ryan@iXsystems.com>
Closes#11191
- Don't leave fstrans set when passed a snapshot
- Don't remove minor if volmode already matches new value
- (FreeBSD) Wait for GEOM ops to complete before trying
remove (at create time GEOM will be "tasting" in parallel)
- (FreeBSD) Don't leak zvol_state_lock on open if zv == NULL
- (FreeBSD) Don't try to unlock zv->zv_state lock if zv == NULL
Reviewed-by: Ryan Moeller <ryan@iXsystems.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Matt Macy <mmacy@FreeBSD.org>
Closes#11199
For encrypted receives, where user accounting is initially disabled on
creation, both 'zfs userspace' and 'zfs groupspace' fails with
EOPNOTSUPP: this is because dmu_objset_id_quota_upgrade_cb() forgets to
set OBJSET_FLAG_USERACCOUNTING_COMPLETE on the objset flags after a
successful dmu_objset_space_upgrade().
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Co-authored-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: loli10K <ezomori.nozomu@gmail.com>
Closes#9501Closes#9596
In case of cache device removal it is possible that at the end of
l2arc_evict() we have l2ad_hand = l2ad_evict. This can lead to the
following panic in case of a debug build:
VERIFY3(dev->l2ad_hand < dev->l2ad_evict) failed (321920512 < 321920512)
Call Trace:
dump_stack+0x66/0x90
spl_panic+0xef/0x117 [spl]
l2arc_remove_vdev+0x11d/0x290 [zfs]
spa_load_l2cache+0x275/0x5b0 [zfs]
spa_vdev_remove+0x4a5/0x6e0 [zfs]
zfs_ioc_vdev_remove+0x59/0xa0 [zfs]
zfsdev_ioctl_common+0x5b3/0x630 [zfs]
zfsdev_ioctl+0x53/0xe0 [zfs]
do_vfs_ioctl+0x42e/0x6b0
ksys_ioctl+0x5e/0x90
do_syscall_64+0x5b/0x1a0
entry_SYSCALL_64_after_hwframe+0x44/0xa9
In case of cache device removal it also possible that l2ad_hand +
distance > l2ad_end since we do not iterate l2arc_evict() and l2ad_hand
is not reset. This has no functional consequence however as the cache
device is about to be removed.
Fix this by omitting the ASSERT in case of device removal.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: George Amanakis <gamanakis@gmail.com>
Closes#11205
The output of ZFS channel programs is logged on-disk in the zpool
history, and printed by `zpool history -i`. Channel programs can use
10MB of memory by default, and up to 100MB by using the `zfs program -m`
flag. Therefore their output can be up to some fraction of 100MB.
In addition to being somewhat wasteful of the limited space reserved for
the pool history (which for large pools is 1GB), in extreme cases this
can result in a failure of `ASSERT(length <= DMU_MAX_ACCESS);` in
`dmu_buf_hold_array_by_dnode()`.
This commit limits the output size that will be logged to 1MB. Larger
outputs will not be logged, instead a entry will be logged indicating
the size of the omitted output.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Matthew Ahrens <mahrens@delphix.com>
Closes#11194
FreeBSD's VFS expects EFAULT from zfs_write() if we didn't complete
the full write so it can retry the operation. Add some missing
SET_ERRORs in zfs_write().
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Ryan Moeller <ryan@iXsystems.com>
Closes#11193
This patch adds a new top-level vdev type called dRAID, which stands
for Distributed parity RAID. This pool configuration allows all dRAID
vdevs to participate when rebuilding to a distributed hot spare device.
This can substantially reduce the total time required to restore full
parity to pool with a failed device.
A dRAID pool can be created using the new top-level `draid` type.
Like `raidz`, the desired redundancy is specified after the type:
`draid[1,2,3]`. No additional information is required to create the
pool and reasonable default values will be chosen based on the number
of child vdevs in the dRAID vdev.
zpool create <pool> draid[1,2,3] <vdevs...>
Unlike raidz, additional optional dRAID configuration values can be
provided as part of the draid type as colon separated values. This
allows administrators to fully specify a layout for either performance
or capacity reasons. The supported options include:
zpool create <pool> \
draid[<parity>][:<data>d][:<children>c][:<spares>s] \
<vdevs...>
- draid[parity] - Parity level (default 1)
- draid[:<data>d] - Data devices per group (default 8)
- draid[:<children>c] - Expected number of child vdevs
- draid[:<spares>s] - Distributed hot spares (default 0)
Abbreviated example `zpool status` output for a 68 disk dRAID pool
with two distributed spares using special allocation classes.
```
pool: tank
state: ONLINE
config:
NAME STATE READ WRITE CKSUM
slag7 ONLINE 0 0 0
draid2:8d:68c:2s-0 ONLINE 0 0 0
L0 ONLINE 0 0 0
L1 ONLINE 0 0 0
...
U25 ONLINE 0 0 0
U26 ONLINE 0 0 0
spare-53 ONLINE 0 0 0
U27 ONLINE 0 0 0
draid2-0-0 ONLINE 0 0 0
U28 ONLINE 0 0 0
U29 ONLINE 0 0 0
...
U42 ONLINE 0 0 0
U43 ONLINE 0 0 0
special
mirror-1 ONLINE 0 0 0
L5 ONLINE 0 0 0
U5 ONLINE 0 0 0
mirror-2 ONLINE 0 0 0
L6 ONLINE 0 0 0
U6 ONLINE 0 0 0
spares
draid2-0-0 INUSE currently in use
draid2-0-1 AVAIL
```
When adding test coverage for the new dRAID vdev type the following
options were added to the ztest command. These options are leverages
by zloop.sh to test a wide range of dRAID configurations.
-K draid|raidz|random - kind of RAID to test
-D <value> - dRAID data drives per group
-S <value> - dRAID distributed hot spares
-R <value> - RAID parity (raidz or dRAID)
The zpool_create, zpool_import, redundancy, replacement and fault
test groups have all been updated provide test coverage for the
dRAID feature.
Co-authored-by: Isaac Huang <he.huang@intel.com>
Co-authored-by: Mark Maybee <mmaybee@cray.com>
Co-authored-by: Don Brady <don.brady@delphix.com>
Co-authored-by: Matthew Ahrens <mahrens@delphix.com>
Co-authored-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Mark Maybee <mmaybee@cray.com>
Reviewed-by: Matt Ahrens <matt@delphix.com>
Reviewed-by: Tony Hutter <hutter2@llnl.gov>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#10102
ZFS channel programs (invoked by `zfs program`) are executed in a LUA
sandbox with a limit on the amount of memory they can consume. The
limit is 10MB by default, and can be raised to 100MB with the `-m` flag.
If the memory limit is exceeded, the LUA program exits and the command
fails with a message like `Channel program execution failed: Memory
limit exhausted.`
The LUA sandbox allocates memory with `vmem_alloc(KM_NOSLEEP)`, which
will fail if the requested memory is not immediately available. In this
case, the program fails with the same message, `Memory limit exhausted`.
However, in this case the specified memory limit has not been reached,
and the memory may only be temporarily unavailable.
This commit changes the LUA memory allocator `zcp_lua_alloc()` to use
`vmem_alloc(KM_SLEEP)`, so that we won't spuriously fail when memory is
temporarily low. Instead, we rely on the system to be able to free up
memory (e.g. by evicting from the ARC), and we assume that even at the
highest memory limit of 100MB, the channel program will not truly
exhaust the system's memory.
External-issue: DLPX-71924
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Ryan Moeller <ryan@iXsystems.com>
Signed-off-by: Matthew Ahrens <mahrens@delphix.com>
Closes#11190
It is a leftover from illumos always set to NULL and introducing a
spurious difference between zio_buf and zio_data_buf.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Mateusz Guzik <mjguzik@gmail.com>
Closes#11188
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Matt Macy <mmacy@FreeBSD.org>
Reviewed-by: Alexander Motin <mav@FreeBSD.org>
Signed-off-by: Ryan Moeller <ryan@iXsystems.com>
Closes#11176
Show that these values will not be changing later.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Matt Macy <mmacy@FreeBSD.org>
Reviewed-by: Alexander Motin <mav@FreeBSD.org>
Signed-off-by: Ryan Moeller <ryan@iXsystems.com>
Closes#11176
The oid comes from the znode we are already passing.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Matt Macy <mmacy@FreeBSD.org>
Reviewed-by: Alexander Motin <mav@FreeBSD.org>
Signed-off-by: Ryan Moeller <ryan@iXsystems.com>
Closes#11176
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Matt Macy <mmacy@FreeBSD.org>
Reviewed-by: Alexander Motin <mav@FreeBSD.org>
Signed-off-by: Ryan Moeller <ryan@iXsystems.com>
Closes#11176
After initial arc_c was reduced to arc_c_min it became possible that
on datasets with primarycache=metadata or none dirty data make up most
of ARC capacity and easily more than configured 50% of initial arc_c,
that causes forced txg commits by arc_tempreserve_space() and periodic
very long write delays.
This patch makes arc_tempreserve_space() to use arc_c only after ARC
warmed up once and arc_c really means something, but use arc_c_max
before that.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Ryan Moeller <ryan@iXsystems.com>
Reviewed-by: Matt Macy <mmacy@FreeBSD.org>
Signed-off-by: Alexander Motin <mav@FreeBSD.org>
Sponsored-By: iXsystems, Inc.
Closes#11178
Fix a couple of places where the wrong tag is passed
to dnode_{hold, rele}
Reviewed-by: Alexander Motin <mav@FreeBSD.org>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Matt Macy <mmacy@FreeBSD.org>
Closes#11184
Move zfs_get_data() in to platform-independent code. The only
platform-specific aspect of it is the way we release an inode
(Linux) / vnode_t (FreeBSD). I am not aware of a platform that
could be supported by ZFS that couldn't implement zfs_rele_async
itself. It's sibling zvol_get_data already is platform-independent.
Reviewed-by: Ryan Moeller <ryan@iXsystems.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Christian Schwarz <me@cschwarz.com>
Closes#10979
Current CPU_SEQID users don't care about possibly changing CPU ID, but
enclose it within kpreempt disable/enable in order to fend off warnings
from Linux's CONFIG_DEBUG_PREEMPT.
There is no need to do it. The expected way to get CPU ID while allowing
for migration is to use raw_smp_processor_id.
In order to make this future-proof this patch keeps CPU_SEQID as is and
introduces CPU_SEQID_UNSTABLE instead, to make it clear that consumers
explicitly want this behavior.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Ryan Moeller <ryan@iXsystems.com>
Reviewed-by: Matt Macy <mmacy@FreeBSD.org>
Signed-off-by: Mateusz Guzik <mjguzik@gmail.com>
Closes#11142
The zfs_holey() and zfs_access() functions can be made common
to both FreeBSD and Linux.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Ryan Moeller <ryan@iXsystems.com>
Signed-off-by: Matt Macy <mmacy@FreeBSD.org>
Closes#11125
The original xuio zero copy functionality has always been unused
on Linux and FreeBSD. Remove this disabled code to avoid any
confusion and improve readability.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Ryan Moeller <ryan@iXsystems.com>
Signed-off-by: Matt Macy <mmacy@FreeBSD.org>
Closes#11124
L2ARC devices of several terabytes filled with 4KB blocks may take 15
minutes to rebuild. Due to the way L2ARC log reading is implemented
it is quite likely that for all that time rebuild thread will never
sleep. At least on FreeBSD kernel threads have absolute priority and
can not be preempted by threads with lower priorities. If some thread
is also bound to that specific CPU it may not get any CPU time for all
the 15 minutes.
Reviewed-by: Cedric Berger <cedric@precidata.com>
Reviewed-by: Ryan Moeller <freqlabs@FreeBSD.org>
Reviewed-by: George Amanakis <gamanakis@gmail.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Alexander Motin <mav@FreeBSD.org>
Closes#11116
