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
Fedora does not guarantee a stable kABI, so weak modules should be dis-
abled. See the dkms man page for a more detailed explanation of the weak
module feature.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Gregory Bartholomew <gregory.lee.bartholomew@gmail.com>
Closes#9891Closes#11128Closes#11242Closes#11335
Avoid a bug with gcc's -Wreturn-local-addr warning with some
obfuscation. In buggy versions of gcc, if a return value is an
expression that involves the address of a local variable, and even if
that address is legally converted to a non-pointer type, a warning may
be emitted and the value of the address may be replaced with zero.
Howerver, buggy versions don't emit the warning or replace the value
when simply returning a local variable of non-pointer type.
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=90737
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Ryan Libby <rlibby@FreeBSD.org>
Closes#11337
Building the spa module for i386 caused gcc to emit
-Wint-to-pointer-cast "cast to pointer from integer of different size"
because spa.spa_did was uint64_t but pthread_join (via thread_join in
spa_deactivate) takes a pointer (32-bit on i386). Define spa_did to be
pointer-size instead. For now spa_did is in fact never non-zero and the
thread_join could instead be ifdef'd out, but changing the size of
spa_did may be more useful for the future.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Ryan Libby <rlibby@FreeBSD.org>
Closes#11336
Building libzfs with gcc on FreeBSD failed because gcc is picky about
the order of keywords in declarations with __thread, whereas clang is
more relaxed.
https://gcc.gnu.org/onlinedocs/gcc/Thread-Local.html
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Ryan Moeller <ryan@iXsystems.com>
Signed-off-by: Ryan Libby <rlibby@FreeBSD.org>
Closes#11331
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
In an earlier revision of dRAID there existed an /etc/zfs/draid.d
directory. This was removed before the final version was integrated
but a little bit was accidentally overlooked in the zfs_helpers.sh
script. Remove this remnant.
Reviewed-by: George Melikov <mail@gmelikov.ru>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#11326
Some tunables shown by arc_summary3 have string values that may exceed
the normal line length, leaving a negative offset between the name and
value fields. The negative space is of course not valid and Python
rightly barfs up an exception traceback.
Handle an overflowing value field width by ignoring the line length
and separating the name from the value by a single space instead.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Ryan Moeller <ryan@iXsystems.com>
Closes#11270
There is a tunable to select the fletcher 4 checksum implementation on
Linux but it was not present in FreeBSD.
Implement the sysctl handler for FreeBSD and use ZFS_MODULE_PARAM_CALL
to provide the tunable on both platforms.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Ryan Moeller <ryan@iXsystems.com>
Closes#11270
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
vfs.zfs.arc_no_grow_shift has an invalid type (15) and this causes
py-sysctl to format it as a bytearray when it should be an integer.
"U" is not a valid format, it should be "I" and the type should match
the variable type, int. We can return EINVAL if the value is set below
zero.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Ryan Moeller <ryan@iXsystems.com>
Closes#11318
py-sysctl now includes the CTLTYPE_NODE type nodes in the list returned
by sysctl.filter() on FreeBSD head. It also provides descriptions now.
Eliminate the subprocess call to get descriptions, and filter out the
nodes so we only deal with values.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Ryan Moeller <ryan@iXsystems.com>
Closes#11318
Resolve an uninitialized variable warning when compiling.
In function ‘zfs_domount’:
warning: ‘root_inode’ may be used uninitialized in this
function [-Wmaybe-uninitialized]
sb->s_root = d_make_root(root_inode);
Reviewed-by: Tony Hutter <hutter2@llnl.gov>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#11306
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
Run ztest via zloop for 20 minutes, total run time is ~30 minutes.
Reviewed-by: Kjeld Schouten <kjeld@schouten-lebbing.nl>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: George Melikov <mail@gmelikov.ru>
Closes#11319
There has been a panic affecting some system configurations where the
thread FPU context is disturbed during the fletcher 4 benchmarks,
leading to a panic at boot.
module_init() registers zcommon_init to run in the last subsystem
(SI_SUB_LAST). Running it as soon as interrupts have been configured
(SI_SUB_INT_CONFIG_HOOKS) makes sure we have finished the benchmarks
before we start doing other things.
While it's not clear *how* the FPU context was being disturbed, this
does seem to avoid it.
Add a module_init_early() macro to run zcommon_init() at this earlier
point on FreeBSD. On Linux this is defined as module_init().
Authored by: Konstantin Belousov <kib@FreeBSD.org>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Ryan Moeller <ryan@iXsystems.com>
Closes#11302
Follow up fix for 0cb40fa3. Remove unused variables, don't source
unused libs and add missed cleanup.
Reviewed-by: Ryan Moeller <ryan@iXsystems.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Attila Fülöp <attila@fueloep.org>
Closes#11311
Tracking down an error message with the errno value can be difficult,
using strerror makes the error message clearer.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Érico Rolim <erico.erc@gmail.com>
Closes#11303
Canonicalization, the source of the trouble, was disabled in 9000a9f.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Sterling Jensen <sterlingjensen@users.noreply.github.com>
Closes#11295
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
Run zfs-tests with sanity.run for brief results. Timeouts
are rare, so minimize false positives by increasing the
default from 60 to 180 seconds.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: George Melikov <mail@gmelikov.ru>
Closes#11304
Otherwise trim may finish before progress checks.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: George Melikov <mail@gmelikov.ru>
Closes#11296
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
With both abd_size and abd_nents being uint_t it makes no sense for
abd_chunkcnt_for_bytes() to return size_t. Random mix of different
types used to count chunks looks bad and makes compiler more difficult
to optimize the code.
In particular on FreeBSD this change allows compiler to completely
optimize out abd_verify_scatter() when built without debug, removing
pointless 64-bit division and even more pointless empty loop.
Reviewed-by: Ryan Moeller <ryan@iXsystems.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Alexander Motin <mav@FreeBSD.org>
Closes#11279
Extend the CI checkstyle workflow to perform the library ABI
checks in the master branch. The intent is not to prevent any
ABI changes but to detect them immediately so when they're
made it's done intentionally.
When the changing the ABI the `make storeabi` target can be
used to generate a new .abi file which can be included with
the commit. This depends on the libabigail utility which is
available from the majority of distribution package managers.
Reviewed-by: George Melikov <mail@gmelikov.ru>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#11287
When running in the CI the zpool_import_012_pos test case occasionally
takes longer than the maximum 600 seconds. When this happens the test
case is considered to have failed but always completes a few minutes
latter. Since the logs suggest nothing has actually failed this commit
increases timeout and removes the exception.
Reviewed-by: George Melikov <mail@gmelikov.ru>
Reviewed-by: Ryan Moeller <ryan@iXsystems.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#11286
Occasionally an out of memory error is hit by this test case
when mounting the filesystems. Try and reduce the likelihood
of this occurring by reducing the thread count from 100 to 50.
It also has the advantage of slightly speeding up the test.
cannot mount 'testpool/testfs3/79': Cannot allocate memory
filesystem successfully created, but not mounted
Reviewed-by: George Melikov <mail@gmelikov.ru>
Reviewed-by: Ryan Moeller <ryan@iXsystems.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#11283
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
`zpool create -n` fails to list cache and spare vdevs.
`zpool add -n` fails to list spare devices.
`zpool split -n` fails to list `special` and `dedup` labels.
`zpool add -n` and `zpool split -n` shouldn't list hole devices.
Reviewed-by: Ryan Moeller <ryan@iXsystems.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Attila Fülöp <attila@fueloep.org>
Closes#11122Closes#11167
Add -u option to 'zfs create' that prevents file system from being
automatically mounted. This is similar to the 'zfs receive -u'.
Authored by: pjd <pjd@FreeBSD.org>
FreeBSD-commit: freebsd/freebsd@35c58230e2
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Allan Jude <allan@klarasystems.com>
Ported-by: Ryan Moeller <ryan@iXsystems.com>
Signed-off-by: Ryan Moeller <ryan@iXsystems.com>
Closes#11254
This run file contains a subset of functional tests which exercise
as much functionality as possible while still executing relatively
quickly. The included tests should take no more than a few seconds
each to run at most. This provides a convenient way to sanity test a
change before committing to a full test run which takes several hours.
$ ./scripts/zfs-tests.sh -r sanity
...
Results Summary
PASS 813
Running Time: 00:14:42
Percent passed: 100.0%
Reviewed-by: Ryan Moeller <ryan@iXsystems.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#11271
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
Busybox's mktemp requires at least six X's in the template, causing
the current sed --in-place check to fail because the file does not
exist. This change adds additional X's to mktemp templates that do
not already have at least six X's in them.
Reviewed-by: Ryan Moeller <ryan@iXsystems.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Quentin Zdanis <zdanisq@gmail.com>
Closes#11269
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
When ZFS_COLOR is set, zpool status shows row headings in bold,
except for the "remove:" heading. This is a quick fix that makes
it print in bold too.
Reviewed-by: Tony Hutter <hutter2@llnl.gov>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Andrew Sun <me@andrewsun.com>
Closes#11255
Move the zpool_influxdb command to /usr/libexec/zfs,
and include the /usr/libexec/zfs path in the system search
directory when running the test suite.
Reviewed-by: Richard Elling <Richard.Elling@RichardElling.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Pavel Snajdr <snajpa@snajpa.net>
Closes#11156Closes#11160Closes#11224
Extend the change made in ae12b02 to verify the zfs kernel
modules are loaded to the rest of the OpenZFS services. If
the modules aren't loaded the neither the share, volume, or
and zed services can be started.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#11243
Despite that dracut has a hard dependency on bash,
its modules doesn't, dracut only has a hard dependency on bash for
module-setup (on a fully usable machine). Inside initramfs, dracut
allows users choose from a list of handful other shells, e.g. bash,
busybox, dash, mkfsh.
In fact, my local machine's initramfs is being built with dash,
and it's functional for a very long time.
Before 64025fa3a (Silence 'make checkbashisms', 2020-08-20), we also
allows our users to have that right, too.
Let's fix the problem 'make checkbashisms' reported and allows our users
to have that right, again.
For 'plymouth' case, let's simply run the command inside the if instead
of checking for the existence of command before running it, because the
status is also failture if plymouth is unavailable.
While we're at it, let's remove an unnecessary fork for grep in
zfs-generator.sh.in and its following complicated 'if elif fi' with
a simple 'case ... esac'.
To support this change, also exclude 90zfs from "make checkbashisms"
because the current CI infrastructure ships an old version of
"checkbashisms", which complains about "command -v", while the current
latest "checkbashisms" thinks it's fine. In the near future, we can
revert that change to "Makefile.am" when CI infrastructure is updated.
Reviewed-by: Gabriel A. Devenyi <gdevenyi@gmail.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Đoàn Trần Công Danh <congdanhqx@gmail.com>
Closes#11244
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
