illumos/illumos-gate@770499e185770499e185https://www.illumos.org/issues/8021
The ARC buf data project (known simply as "ABD" since its genesis in the ZoL
community) changes the way the ARC allocates `b_pdata` memory from using linear
`void *` buffers to using scatter/gather lists of fixed-size 1KB chunks. This
improves ZFS's performance by helping to defragment the address space occupied
by the ARC, in particular for cases where compressed ARC is enabled. It could
also ease future work to allocate pages directly from `segkpm` for minimal-
overhead memory allocations, bypassing the `kmem` subsystem.
This is essentially the same change as the one which recently landed in ZFS on
Linux, although they made some platform-specific changes while adapting this
work to their codebase:
1. Implemented the equivalent of the `segkpm` suggestion for future work
mentioned above to bypass issues that they've had with the Linux kernel memory
allocator.
2. Changed the internal representation of the ABD's scatter/gather list so it
could be used to pass I/O directly into Linux block device drivers. (This
feature is not available in the illumos block device interface yet.)
FreeBSD notes:
- the actual (default) chunk size is 4KB (despite the text above saying 1KB)
- we can try to reimplement ABDs, so that they are not permanently
mapped into the KVA unless explicitly requested, especially on
platforms with scarce KVA
- we can try to use unmapped I/O and avoid intermediate allocation of a
linear, virtual memory mapped buffer
- we can try to avoid extra data copying by referring to chunks / pages
in the original ABD
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Reviewed by: George Wilson <george.wilson@delphix.com>
Reviewed by: Paul Dagnelie <pcd@delphix.com>
Reviewed by: John Kennedy <john.kennedy@delphix.com>
Reviewed by: Prakash Surya <prakash.surya@delphix.com>
Reviewed by: Prashanth Sreenivasa <pks@delphix.com>
Reviewed by: Pavel Zakharov <pavel.zakharov@delphix.com>
Reviewed by: Chris Williamson <chris.williamson@delphix.com>
Approved by: Richard Lowe <richlowe@richlowe.net>
Author: Dan Kimmel <dan.kimmel@delphix.com>
MFC after: 3 weeks
illumos/illumos-gate@dbfd9f9300dbfd9f9300https://www.illumos.org/issues/8156
dbuf_evict_notify() holds the dbuf_evict_lock while checking if it should do
the eviction itself (because the evict thread is not able to keep up).
This can result in massive lock contention.
It isn't necessary to hold the lock, because if we make the wrong choice
occasionally, nothing bad will happen.
Reviewed by: Dan Kimmel <dan.kimmel@delphix.com>
Reviewed by: Paul Dagnelie <pcd@delphix.com>
Approved by: Robert Mustacchi <rm@joyent.com>
Author: Matthew Ahrens <mahrens@delphix.com>
MFC after: 1 week
illumos/illumos-gate@adaec86ad2adaec86ad2https://www.illumos.org/issues/8155
When writing pre-compressed buffers, arc_write() requires that the compression
algorithm used to compress the buffer matches the compression algorithm
requested by the zio_prop_t, which is set by dmu_write_policy().
This makes dmu_write_policy() and its callers a bit more complicated.
We can simplify this by making arc_write() trust the caller to supply the type
of pre-compressed buffer that it wants to write, and override the compression
setting in the zio_prop_t.
Reviewed by: Dan Kimmel <dan.kimmel@delphix.com>
Reviewed by: George Wilson <george.wilson@delphix.com>
Approved by: Robert Mustacchi <rm@joyent.com>
Author: Matthew Ahrens <mahrens@delphix.com>
MFC after: 10 days
illumos/illumos-gate@40713f2b2440713f2b24https://www.illumos.org/issues/8070
Add some ZFS comments left by various developers at different times
Reviewed by: Yuri Pankov <yuri.pankov@gmail.com>
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Approved by: Robert Mustacchi <rm@joyent.com>
Author: Alan Somers <asomers@gmail.com>
MFC after: 1 week
illumos/illumos-gate@def4fac588def4fac588https://www.illumos.org/issues/8025
dbuf_read() creates a zio_root() to track and wait for all the zio's
that may happen as part of this call. However, if the blkptr_t for
this buffer is NULL or a hole, we will not create any more zio's, so
this zio_root() is unnecessary. This is always the case when calling
dbuf_read() on a bonus buffer, because it has no blkptr (it's part of
the containing dnode). For workloads that read a lot of bonus buffers
(e.g. file creation and removal), creating and destroying these
unnecessary zio's can decrease performance by around 3%.
Reviewed by: Dan Kimmel <dan.kimmel@delphix.com>
Reviewed by: Pavel Zakharov <pavel.zakharov@delphix.com>
Reviewed by: Prashanth Sreenivasa <pks@delphix.com>
Approved by: Robert Mustacchi <rm@joyent.com>
Author: Matthew Ahrens <mahrens@delphix.com>
illumos/illumos-gate@94c2d0eb2294c2d0eb22https://www.illumos.org/issues/7968
spa_sync() iterates over all the dirty dnodes and processes each of them by
calling dnode_sync(). If there are many dirty dnodes (e.g. because we created
or removed a lot of files), the single thread of spa_sync() calling
dnode_sync() can become a bottleneck. Additionally, if many dnodes are dirtied
concurrently in open context (e.g. due to concurrent file creation), the
os_lock will experience lock contention via dnode_setdirty().
The solution is to track dirty dnodes on a multilist_t, and for spa_sync() to
use separate threads to process each of the sublists in the multilist.
On the concurrent file creation microbenchmark, the performance improvement
from dnode_setdirty() is up to 7%. Additionally, the wall clock time spent in
spa_sync() is reduced to 15%-40% of the single-threaded case. In terms of cost/
reward, once the other bottlenecks are addressed, fixing this bug will provide
a medium-large performance gain and require a medium amount of effort to
implement.
Reviewed by: Pavel Zakharov <pavel.zakharov@delphix.com>
Reviewed by: Brad Lewis <brad.lewis@delphix.com>
Reviewed by: Saso Kiselkov <saso.kiselkov@nexenta.com>
Reviewed by: Brian Behlendorf <behlendorf1@llnl.gov>
Approved by: Dan McDonald <danmcd@omniti.com>
Author: Matthew Ahrens <mahrens@delphix.com>
MFC after: 3 weeks
illumos/illumos-gate@10fbdecb0510fbdecb05https://www.illumos.org/issues/7970
The global tunable zfs_arc_num_sublists_per_state is used by the ARC and
the dbuf cache, and other users are planned. We should change this
tunable to be common to all multilists.
Reviewed by: Pavel Zakharov <pavel.zakharov@delphix.com>
Reviewed by: Brad Lewis <brad.lewis@delphix.com>
Reviewed by: Saso Kiselkov <saso.kiselkov@nexenta.com>
Reviewed by: Brian Behlendorf <behlendorf1@llnl.gov>
Approved by: Dan McDonald <danmcd@omniti.com>
Author: Matthew Ahrens <mahrens@delphix.com>
MFC after: 3 weeks
illumos/illumos-gate@61e255ce7261e255ce72https://www.illumos.org/issues/7793
Background information: This assertion about tx_space_* verifies that we
are not dirtying more stuff than we thought we would. We “need” to know
how much we will dirty so that we can check if we should fail this
transaction with ENOSPC/EDQUOT, in dmu_tx_assign(). While the
transaction is open (i.e. between dmu_tx_assign() and dmu_tx_commit() —
typically less than a millisecond), we call dbuf_dirty() on the exact
blocks that will be modified. Once this happens, the temporary
accounting in tx_space_* is unnecessary, because we know exactly what
blocks are newly dirtied; we call dnode_willuse_space() to track this
more exact accounting.
The fundamental problem causing this bug is that dmu_tx_hold_*() relies
on the current state in the DMU (e.g. dn_nlevels) to predict how much
will be dirtied by this transaction, but this state can change before we
actually perform the transaction (i.e. call dbuf_dirty()).
This bug will be fixed by removing the assertion that the tx_space_*
accounting is perfectly accurate (i.e. we never dirty more than was
predicted by dmu_tx_hold_*()). By removing the requirement that this
accounting be perfectly accurate, we can also vastly simplify it, e.g.
removing most of the logic in dmu_tx_count_*().
The new tx space accounting will be very approximate, and may be more or
less than what is actually dirtied. It will still be used to determine
if this transaction will put us over quota. Transactions that are marked
by dmu_tx_mark_netfree() will be excepted from this check. We won’t make
an attempt to determine how much space will be freed by the transaction
— this was rarely accurate enough to determine if a transaction should
be permitted when we are over quota, which is why dmu_tx_mark_netfree()
was introduced in 2014.
We also won’t attempt to give “credit” when overwriting existing blocks,
if those blocks may be freed. This allows us to remove the
do_free_accounting logic in dbuf_dirty(), and associated routines. This
Reviewed by: Steve Gonczi <steve.gonczi@delphix.com>
Reviewed by: George Wilson <george.wilson@delphix.com>
Reviewed by: Pavel Zakharov <pavel.zakharov@delphix.com>
Reviewed by: Brian Behlendorf <behlendorf1@llnl.gov>
Approved by: Robert Mustacchi <rm@joyent.com>
Author: Matthew Ahrens <mahrens@delphix.com>
MFC after: 3 weeks
7740 fix for 6513 only works in hole punching case, not truncation
illumos/illumos-gate@7de35a3ed07de35a3ed0https://www.illumos.org/issues/7740
The problem is that dbuf_findbp will return ENOENT if the block it's
trying to find is beyond the end of the file. If that happens, we assume
there is no birth time, and so we lose that information when we write
out new blkptrs. We should teach dbuf_findbp to look for things that are
beyond the current end, but not beyond the absolute end of the file.
To verify, create a large file, truncate it to a short length, and then
write beyond the end. Check with zdb to make sure that there are no
holes with birth time zero (will appear as gaps).
Reviewed by: Steve Gonczi <steve.gonczi@delphix.com>
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Approved by: Dan McDonald <danmcd@omniti.com>
Author: Paul Dagnelie <pcd@delphix.com>
7580 ztest failure in dbuf_read_impl
illumos/illumos-gate@1a01181fdc1a01181fdchttps://www.illumos.org/issues/7580
We need to prevent any reader whenever we're about the zero out all the
blkptrs. To do this we need to grab the dn_struct_rwlock as writer in
dbuf_write_children_ready and free_children just prior to calling bzero.
Reviewed by: Pavel Zakharov <pavel.zakharov@delphix.com>
Reviewed by: Steve Gonczi <steve.gonczi@delphix.com>
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Approved by: Dan McDonald <danmcd@omniti.com>
Author: George Wilson <george.wilson@delphix.com>
7252 7628 compressed zfs send / receive
illumos/illumos-gate@5602294fda5602294fdahttps://www.illumos.org/issues/7252
This feature includes code to allow a system with compressed ARC enabled to
send data in its compressed form straight out of the ARC, and receive data in
its compressed form directly into the ARC.
https://www.illumos.org/issues/7628
We should have longer, more readable versions of the ZFS send / recv options.
7628 create long versions of ZFS send / receive options
Reviewed by: George Wilson <george.wilson@delphix.com>
Reviewed by: John Kennedy <john.kennedy@delphix.com>
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Reviewed by: Paul Dagnelie <pcd@delphix.com>
Reviewed by: Pavel Zakharov <pavel.zakharov@delphix.com>
Reviewed by: Sebastien Roy <sebastien.roy@delphix.com>
Reviewed by: David Quigley <dpquigl@davequigley.com>
Reviewed by: Thomas Caputi <tcaputi@datto.com>
Approved by: Dan McDonald <danmcd@omniti.com>
Author: Dan Kimmel <dan.kimmel@delphix.com>
6676 Race between unique_insert() and unique_remove() causes ZFS fsid change
illumos/illumos-gate@40510e8eba40510e8ebahttps://www.illumos.org/issues/6676
The fsid of zfs filesystems might change after reboot or remount. The problem seems to
be caused by a race between unique_insert() and unique_remove(). The unique_remove()
is called from dsl_dataset_evict() which is now an asynchronous thread. In a case the
dsl_dataset_evict() thread is very slow and calls unique_remove() too late we will end
up with changed fsid on zfs mount.
This problem is very likely caused by #5056.
Steps to Reproduce
Note: I'm able to reproduce this always on a single core (virtual) machine. On multicore
machines it is not so easy to reproduce.
# uname -a
SunOS openindiana 5.11 illumos-633aa80 i86pc i386 i86pc Solaris
# zfs create rpool/TEST
# FS=$(echo ::fsinfo | mdb -k | grep TEST | awk '{print $1}')
# echo $FS::print vfs_t vfs_fsid | mdb -k
vfs_fsid = {
vfs_fsid.val = [ 0x54d7028a, 0x70311508 ]
}
# zfs umount rpool/TEST
# zfs mount rpool/TEST
# FS=$(echo ::fsinfo | mdb -k | grep TEST | awk '{print $1}')
# echo $FS::print vfs_t vfs_fsid | mdb -k
vfs_fsid = {
vfs_fsid.val = [ 0xd9454e49, 0x6b36d08 ]
}
#
Impact
The persistent fsid (filesystem id) is essential for proper NFS functionality.
If the fsid of a filesystem changes on remount (or after reboot) the NFS
clients might not be able to automatically recover from such event and the
manual remount of the NFS filesystems on every NFS client might be needed.
Author: Josef 'Jeff' Sipek <josef.sipek@nexenta.com>
Reviewed by: Saso Kiselkov <saso.kiselkov@nexenta.com>
Reviewed by: Sanjay Nadkarni <sanjay.nadkarni@nexenta.com>
Reviewed by: Dan Vatca <dan.vatca@gmail.com>
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Reviewed by: George Wilson <george.wilson@delphix.com>
Reviewed by: Sebastien Roy <sebastien.roy@delphix.com>
Approved by: Robert Mustacchi <rm@joyent.com>
7500 Simplify dbuf_free_range by removing dn_unlisted_l0_blkid
illumos/illumos-gate@653af1b809653af1b809https://www.illumos.org/issues/7500
With the integration of:
commit 0f6d88aded0d165f5954688a9b13bac76c38da84
Author: Alex Reece <alex@delphix.com>
Date: Sat Jul 26 13:40:04 2014 -0800
4873 zvol unmap calls can take a very long time for larger datasets
the dnode's dn_bufs field was changed from a list to a tree. As a result,
the dn_unlisted_l0_blkid field is no longer necessary.
Author: Stephen Blinick <stephen.blinick@delphix.com>
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Reviewed by: Dan Kimmel <dan.kimmel@delphix.com>
Approved by: Gordon Ross <gordon.w.ross@gmail.com>
dsl_dataset_space is looking at the ds_bp's fill count while
dmu_objset_write_ready() is concurrently modifying it. This fix adds an
rrwlock to protect the ds_bp.
Closes#180
Reviewed by: George Wilson <george.wilson@delphix.com>
Reviewed by: Prakash Surya <prakash.surya@delphix.com>
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Reviewed by: Steve Gonczi <steve.gonczi@delphix.com>
Author: Paul Dagnelie <pcd@delphix.com>
Using a benchmark which has 32 threads creating 2 million files in the
same directory, on a machine with 16 CPU cores, I observed poor
performance. I noticed that dmu_tx_hold_zap() was using about 30% of
all CPU, and doing dnode_hold() 7 times on the same object (the ZAP
object that is being held).
dmu_tx_hold_zap() keeps a hold on the dnode_t the entire time it is
running, in dmu_tx_hold_t:txh_dnode, so it would be nice to use the
dnode_t that we already have in hand, rather than repeatedly calling
dnode_hold(). To do this, we need to pass the dnode_t down through
all the intermediate calls that dmu_tx_hold_zap() makes, making these
routines take the dnode_t* rather than an objset_t* and a uint64_t
object number. In particular, the following routines will need to have
analogous *_by_dnode() variants created:
dmu_buf_hold_noread()
dmu_buf_hold()
zap_lookup()
zap_lookup_norm()
zap_count_write()
zap_lockdir()
zap_count_write()
This can improve performance on the benchmark described above by 100%,
from 30,000 file creations per second to 60,000. (This improvement is on
top of that provided by working around the object allocation issue. Peak
performance of ~90,000 creations per second was observed with 8 CPUs;
adding CPUs past that decreased performance due to lock contention.) The
CPU used by dmu_tx_hold_zap() was reduced by 88%, from 340 CPU-seconds
to 40 CPU-seconds.
Sponsored by: Intel Corp.
Closes#109
Reviewed by: Steve Gonczi <steve.gonczi@delphix.com>
Reviewed by: George Wilson <george.wilson@delphix.com>
Reviewed by: Pavel Zakharov <pavel.zakharov@delphix.com>
Reviewed by: Ned Bass <bass6@llnl.gov>
Reviewed by: Brian Behlendorf <behlendorf1@llnl.gov>
Author: Matthew Ahrens <mahrens@delphix.com>
openzfs/openzfs@d3e523d489
zap_lockdir() / zap_unlockdir() should take a "void *tag" argument which
tags the hold on the zap. This will help diagnose programming errors
which misuse the hold on the ZAP.
Sponsored by: Intel Corp.
Closes#108
Reviewed by: Pavel Zakharov <pavel.zakharov@delphix.com>
Reviewed by: Steve Gonczi <steve.gonczi@delphix.com>
Reviewed by: George Wilson <george.wilson@delphix.com>
Reviewed by: Brian Behlendorf <behlendorf1@llnl.gov>
Author: Matthew Ahrens <mahrens@delphix.com>
openzfs/openzfs@0780b3eab5
7086 ztest attempts dva_get_dsize_sync on an embedded blockpointer
illumos/illumos-gate@926549256bhttps://github.com/illumos/illumos-gate/commit/926549256b71acd595f69b236779ff6b7
8fa08ef
https://www.illumos.org/issues/7086
In dbuf_dirty(), we need to grab the dn_struct_rwlock before looking at the
db_blkptr, to prevent it from being changed by syncing context.
Otherwise we may see that ztest got a segfault from this stack:
libzpool.so.1`dva_get_dsize_sync+0x98(872f000, b32b240, fed7811b, 0, b4cda20,
0)
libzpool.so.1`bp_get_dsize+0x60(872f000, b32b240, 0, 97cb780, 9d4c1a8, 0)
libzpool.so.1`dbuf_dirty+0x9b3(ce0a100, 97cb780, 9, fecd2530)
libzpool.so.1`dmu_buf_will_dirty+0xc3(ce0a100, 97cb780, ea293d6c, 1)
libzpool.so.1`zap_lockdir+0x1a0(8aaa3c0, 1, 0, 97cb780, 1, 1)
libzpool.so.1`zap_remove_norm+0x30(8aaa3c0, 1, 0, 8728b10, 0, 97cb780)
libzpool.so.1`zap_remove+0x29(8aaa3c0, 1, 0, 8728b10, 97cb780, a)
ztest_replay_remove+0x225(ea294588, 8728ae8, 0, 38010000, 0, 0)
ztest_remove+0x9f(ea294588, ea293f50, 4, 3)
ztest_object_init+0x78(ea294588, ea293f50, 4e0, 1)
ztest_dmu_object_alloc_free+0x71(ea294588, 13)
ztest_dmu_objset_create_destroy+0x224(80cef08, 13, 0, 805d36c, 9017ad44, 0)
ztest_execute+0x89(a, 807c720, 13, 0)
ztest_thread+0xea(13, 0, 0, 0)
libc.so.1`_thrp_setup+0x88(f0983240)
libc.so.1`_lwp_start(f0983240, 0, 0, 0, 0, 0)
Looking into it a bit, we see that this is an embedded blockpointer, so
BP_GET_NDVAS should have returned 0:
b32b240::blkptr
EMBEDDED [L0 ZAP_OTHER] et=0 LZ4 size=200L/4aP birth=80L
Instead, it looks like another thread is modifying this blockpointer:
b32b240::ugrep | ::whatis
f47a0e0c is in [ stack tid=0x19f ]
ebd6ec40 is in [ stack tid=0x226 ]
ea293bd0 is in [ stack tid=0x244 ]
ea293be4 is in [ stack tid=0x244 ]
Reviewed by: Prakash Surya <prakash.surya@delphix.com>
Reviewed by: George Wilson <george.wilson@delphix.com>
Approved by: Robert Mustacchi <rm@joyent.com>
Author: Matthew Ahrens <mahrens@delphix.com>
illumos/illumos-gate@dcbf3bd6a1dcbf3bd6a1https://www.illumos.org/issues/6950
When reading compressed data from disk, the ARC should keep the compressed
block cached and only decompress it when consumers access the block. The
uncompressed data should be short-lived allowing the ARC to cache a much larger
amount of data. The DMU would also maintain a smaller cache of uncompressed
blocks to minimize the impact of decompressing frequently accessed blocks.
Reviewed by: Prakash Surya <prakash.surya@delphix.com>
Reviewed by: Dan Kimmel <dan.kimmel@delphix.com>
Reviewed by: Matt Ahrens <mahrens@delphix.com>
Reviewed by: Paul Dagnelie <pcd@delphix.com>
Reviewed by: Don Brady <don.brady@intel.com>
Reviewed by: Richard Elling <Richard.Elling@RichardElling.com>
Approved by: Richard Lowe <richlowe@richlowe.net>
Author: George Wilson <george.wilson@delphix.com>
illumos/illumos-gate@8df0bcf0df8df0bcf0dfhttps://www.illumos.org/issues/6513
If a ZFS object contains a hole at level one, and then a data block is created
at level 0 underneath that l1 block, l0 holes will be created. However, these
l0 holes do not have the birth time property set; as a result, incremental
sends will not send those holes.
Fix is to modify the dbuf_read code to fill in birth time data.
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Reviewed by: George Wilson <george.wilson@delphix.com>
Reviewed by: Boris Protopopov <bprotopopov@hotmail.com>
Approved by: Richard Lowe <richlowe@richlowe.net>
Author: Paul Dagnelie <pcd@delphix.com>
MFC after: 3 weeks
illumos/illumos-gate@11ceac77ea11ceac77eahttps://www.illumos.org/issues/6844
dnode_next_offset is used in a variety of places to iterate over the holes or
allocated blocks in a dnode. It operates under the premise that it can iterate
over the blockpointers of a dnode in open context while holding only the
dn_struct_rwlock as reader. Unfortunately, this premise does not hold.
When we create the zio for a dbuf, we pass in the actual block pointer in the
indirect block above that dbuf. When we later zero the bp in
zio_write_compress, we are directly modifying the bp. The state of the bp is
now inconsistent from the perspective of dnode_next_offset: the bp will appear
to be a hole until zio_dva_allocate finally finishes filling it in. In the
meantime, dnode_next_offset can detect a hole in the dnode when none exists.
I was able to experimentally demonstrate this behavior with the following
setup:
1. Create a file with 1 million dbufs.
2. Create a thread that randomly dirties L2 blocks by writing to the first L0
block under them.
3. Observe dnode_next_offset, waiting for it to skip over a hole in the middle
of a file.
4. Do dnode_next_offset in a loop until we skip over such a non-existent hole.
The fix is to ensure that it is valid to iterate over the indirect blocks in a
dnode while holding the dn_struct_rwlock by passing the zio a copy of the BP
and updating the actual BP in dbuf_write_ready while holding the lock.
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Reviewed by: George Wilson <george.wilson@delphix.com>
Reviewed by: Boris Protopopov <bprotopopov@hotmail.com>
Approved by: Dan McDonald <danmcd@omniti.com>
Author: Alex Reece <alex@delphix.com>
MFC after: 3 weeks
Reviewed by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed by: Dan McDonald <danmcd@omniti.com>
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Approved by: Robert Mustacchi <rm@joyent.com>
Author: Tim Chase <tim@chase2k.com>
openzfs/openzfs@445e67805d
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Reviewed by: Paul Dagnelie <pcd@delphix.com>
Author: Alexander Motin <mav@FreeBSD.org>
Improve speculative prefetch of indirect blocks.
Scalability of many operations on wide ZFS pool can be limited by
requirement to prefetch indirect blocks first. Recently added
asynchronous indirect block read partially helped, but did not
solve the problem completely. This patch extends existing prefetcher
functionality to explicitly work with indirect blocks.
Before this change prefetcher issued reads for up to 8MB of data in
advance. With this change it also issues indirect block reads
for up to 64MB of data in advance, so that when it will be time to
actually read those data, it can be done immediately. Alike effect
can be achieved by just increasing maximal data prefetch distance,
but at higher memory cost.
Also this change introduces indirect block prefetch for rewrite
operations, that was never done before. Previously ARC miss for
Indirect blocks regularly blocked rewrites, converting perfectly
aligned asynchronous operations into synchronous read-write pairs,
significantly reducing maximal rewrite speed.
While being there this issue was also fixed:
- prefetch was done always, even if caching for the dataset was
completely disabled.
Testing on FreeBSD with zvol on top of 6x striped 2x mirrored pool
of 12 assorted HDDs shown me such performance numbers:
------- BEFORE --------
Write 491363677 bytes/sec
Read 312430631 bytes/sec
Rewrite 97680464 bytes/sec
-------- AFTER --------
Write 493524146 bytes/sec
Read 438598079 bytes/sec
Rewrite 277506044 bytes/sec
Closes#65Closes#80openzfs/openzfs@792fd28ac0
Reviewed by: Richard Yao <ryao@gentoo.org>
Reviewed by: Xin LI <delphij@freebsd.org>
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Approved by: Richard Lowe <richlowe@richlowe.net>
Author: Justin T. Gibbs <gibbs@FreeBSD.org>
illumos/illumos-gate@d2058105c6
Reviewed by: George Wilson <george.wilson@delphix.com>
Reviewed by: Paul Dagnelie <pcd@delphix.com>
Reviewed by: Justin Gibbs <gibbs@scsiguy.com>
Reviewed by: Richard Elling <Richard.Elling@RichardElling.com>
Approved by: Robert Mustacchi <rm@joyent.com>
Author: Matthew Ahrens <mahrens@delphix.com>
illumos/illumos-gate@0f2e7d03b8
Rewrite the ZFS prefetch code to detect only forward, sequential
streams.
The following kstats have been added:
kstat.zfs.misc.arcstats.sync_wait_for_async
How many sync reads have waited for async read
to complete. (less is better)
kstat.zfs.misc.arcstats.demand_hit_predictive_prefetch
How many demand read didn't have to wait for I/O
because of predictive prefetch. (more is better)
zfetch kstats have been similified to hits, misses, and max_streams,
with max_streams representing times when we were not able to create
new stream because we already have the maximum number of sequences
for a file.
The sysctl variable/loader tunable vfs.zfs.zfetch.block_cap have been
replaced by vfs.zfs.zfetch.max_distance, which controls maximum bytes
to prefetch per stream.
illumos/illumos-gate@cf6106c8a0
Illumos ZFS issues:
5987 zfs prefetch code needs work
https://www.illumos.org/issues/5987
Reviewed by: Toomas Soome <tsoome@me.com>
Reviewed by: George Wilson <george@delphix.com>
Reviewed by: Alex Reece <alex@delphix.com>
Approved by: Richard Lowe <richlowe@richlowe.net>
Author: Matthew Ahrens <mahrens@delphix.com>
illumos/illumos-gate@ca0cc3918a
A ZFS feature flags (large blocks) tracks its refcounts as the number of
datasets that have ever used the feature. Several features of this type
are planned to be added (new checksum functions). This code should be made
common infrastructure rather than duplicating the code for each feature.
5925 zfs receive -o origin=
Reviewed by: Prakash Surya <prakash.surya@delphix.com>
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Author: Paul Dagnelie <pcd@delphix.com>
While running 'zfs recv' we noticed that every 128th 8K block required a
read. We were seeing that restore_write() was calling dmu_tx_hold_write()
and the indirect block was not cached. We should prefetch upcoming indirect
blocks to avoid having to go to disk and blocking the restore_write().
Allow an incremental send stream to be received as a clone, even if the
stream does not mark it as a clone.
Reviewed by: Will Andrews <willa@spectralogic.com>
Reviewed by: Matt Ahrens <mahrens@delphix.com>
Reviewed by: George Wilson <george.wilson@delphix.com>
Approved by: Dan McDonald <danmcd@omniti.com>
Author: Justin Gibbs <justing@spectralogic.com>
illumos/illumos-gate@bc9014e6a8
5630 stale bonus buffer in recycled dnode_t leads to data corruption
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Reviewed by: George Wilson <george@delphix.com>
Reviewed by: Will Andrews <will@freebsd.org>
Approved by: Robert Mustacchi <rm@joyent.com>
Author: Justin T. Gibbs <justing@spectralogic.com>
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Reviewed by: Dan McDonald <danmcd@omniti.com>
Reviewed by: George Wilson <george@delphix.com>
Reviewed by: Bayard Bell <buffer.g.overflow@gmail.com>
Approved by: Robert Mustacchi <rm@joyent.com>
Author: Justin T. Gibbs <justing@spectralogic.com>
illumos/illumos-gate@e57a022b8f
Reviewed by: Bayard Bell <buffer.g.overflow@gmail.com>
Reviewed by: Alek Pinchuk <alek@nexenta.com>
Reviewed by: Simon Klinkert <simon.klinkert@gmail.com>
Reviewed by: Dan McDonald <danmcd@omniti.com>
Approved by: Richard Lowe <richlowe@richlowe.net>
Author: Matthew Ahrens <mahrens@delphix.com>
illumos/illumos-gate@46e1baa6cfhttps://www.illumos.org/issues/5911
Sometimes ZFS appears to hang while deleting a file. It is actually
making slow progress at the file deletion, but other operations
(administrative and writes via the data path) "hang" until the file
removal completes, which can take a long time if the file has many
blocks. The deletion (or most of it) happens in a single txg, and the
sync thread spends most of its time reading indirect blocks via this
stack trace:
swtch+0x141()
cv_wait+0x70()
zio_wait+0x5b()
dbuf_read+0x2c0()
free_children+0x50()
free_children+0x12a()
free_children+0x12a()
free_children+0x12a()
dnode_sync_free_range_impl+0xdf()
dnode_sync_free_range+0x52()
range_tree_vacate+0x65()
dnode_sync+0x1d8()
dmu_objset_sync_dnodes+0x77()
dmu_objset_sync+0x19f()
dsl_dataset_sync+0x51()
dsl_pool_sync+0x9a()
spa_sync+0x2ff()
txg_sync_thread+0x21f()
thread_start+8()
One way to reproduce the problem is if we are over the arc_meta_limit,
e.g. because lots of indirect blocks are pinned because we have L0
dbufs under them. It could be that most of the L1 indirects are cached,
in which case when dmu_free_long_range_impl() calls dmu_tx_hold_free(),
it will complete very quickly. This allows dmu_free_long_range_impl() to
put many (perhaps all of its) transactions in the same TXG. However,
dmu_free_long_range_impl() calls dnode_evict_dbufs (and
dnode_free_range()), which removes the L0 dbufs, thus reducing the hold
count on the L1 indirect blocks above it, allowing them to be evicted.
Because we are over the arc_meta_limit(), these L1 blocks will be
evicted ASAP. Thus when we get to syncing context, the L1 indirects are
no longer cached and must be read in.
Obtained from: illumos
MFC after: 15 days
Convert ARC flags to use enum. Previously, public flags are defined in
arc.h and private flags are defined in arc.c which can lead to confusion
and programming errors.
Consistently use 'hdr' (when referencing arc_buf_hdr_t) instead of 'buf'
or 'ab' because arc_buf_t are often named 'buf' as well.
Illumos issue:
5369 arc flags should be an enum
5370 consistent arc_buf_hdr_t naming scheme
MFC after: 2 weeks
Remove "dbuf phys" db->db_data pointer aliases.
Use function accessors that cast db->db_data to the appropriate
"phys" type, removing the need for clients of the dmu buf user
API to keep properly typed pointer aliases to db->db_data in order
to conveniently access their data.
sys/cddl/contrib/opensolaris/uts/common/fs/zfs/zap_leaf.c:
In zap_leaf() and zap_leaf_byteswap, now that the pointer alias
field l_phys has been removed, use the db_data field in an on
stack dmu_buf_t to point to the leaf's phys data.
sys/cddl/contrib/opensolaris/uts/common/fs/zfs/dbuf.c:
Remove the db_user_data_ptr_ptr field from dbuf and all logic
to maintain it.
sys/cddl/contrib/opensolaris/uts/common/fs/zfs/dbuf.c:
sys/cddl/contrib/opensolaris/uts/common/fs/zfs/dnode.c:
sys/cddl/contrib/opensolaris/uts/common/fs/zfs/sys/dbuf.h:
sys/cddl/contrib/opensolaris/uts/common/fs/zfs/sys/dmu.h:
sys/cddl/contrib/opensolaris/uts/common/fs/zfs/dsl_dataset.c:
sys/cddl/contrib/opensolaris/uts/common/fs/zfs/dsl_dir.c:
sys/cddl/contrib/opensolaris/uts/common/fs/zfs/sa.c:
sys/cddl/contrib/opensolaris/uts/common/fs/zfs/zap.c:
sys/cddl/contrib/opensolaris/uts/common/fs/zfs/zap_micro.c:
Modify the DMU buf user API to remove the ability to specify
a db_data aliasing pointer (db_user_data_ptr_ptr).
cddl/contrib/opensolaris/cmd/zdb/zdb.c:
sys/cddl/contrib/opensolaris/uts/common/fs/zfs/dmu_diff.c:
sys/cddl/contrib/opensolaris/uts/common/fs/zfs/dmu_objset.c:
sys/cddl/contrib/opensolaris/uts/common/fs/zfs/dmu_send.c:
sys/cddl/contrib/opensolaris/uts/common/fs/zfs/dmu_traverse.c:
sys/cddl/contrib/opensolaris/uts/common/fs/zfs/dmu_tx.c:
sys/cddl/contrib/opensolaris/uts/common/fs/zfs/dsl_bookmark.c:
sys/cddl/contrib/opensolaris/uts/common/fs/zfs/dsl_dataset.c:
sys/cddl/contrib/opensolaris/uts/common/fs/zfs/dsl_deadlist.c:
sys/cddl/contrib/opensolaris/uts/common/fs/zfs/dsl_deleg.c:
sys/cddl/contrib/opensolaris/uts/common/fs/zfs/dsl_destroy.c:
sys/cddl/contrib/opensolaris/uts/common/fs/zfs/dsl_dir.c:
sys/cddl/contrib/opensolaris/uts/common/fs/zfs/dsl_pool.c:
sys/cddl/contrib/opensolaris/uts/common/fs/zfs/dsl_prop.c:
sys/cddl/contrib/opensolaris/uts/common/fs/zfs/dsl_scan.c:
sys/cddl/contrib/opensolaris/uts/common/fs/zfs/dsl_synctask.c:
sys/cddl/contrib/opensolaris/uts/common/fs/zfs/dsl_userhold.c:
sys/cddl/contrib/opensolaris/uts/common/fs/zfs/sa.c:
sys/cddl/contrib/opensolaris/uts/common/fs/zfs/spa.c:
sys/cddl/contrib/opensolaris/uts/common/fs/zfs/spa_history.c:
sys/cddl/contrib/opensolaris/uts/common/fs/zfs/zap.c:
sys/cddl/contrib/opensolaris/uts/common/fs/zfs/zap_leaf.c:
sys/cddl/contrib/opensolaris/uts/common/fs/zfs/zap_micro.c:
sys/cddl/contrib/opensolaris/uts/common/fs/zfs/zfs_ioctl.c:
sys/cddl/contrib/opensolaris/uts/common/fs/zfs/sys/dsl_dataset.h:
sys/cddl/contrib/opensolaris/uts/common/fs/zfs/sys/dsl_dir.h:
sys/cddl/contrib/opensolaris/uts/common/fs/zfs/sys/zap_impl.h:
sys/cddl/contrib/opensolaris/uts/common/fs/zfs/sys/zap_leaf.h:
Create and use the new "phys data" accessor functions
dsl_dir_phys(), dsl_dataset_phys(), zap_m_phys(),
zap_f_phys(), and zap_leaf_phys().
sys/cddl/contrib/opensolaris/uts/common/fs/zfs/sys/dsl_dataset.h:
sys/cddl/contrib/opensolaris/uts/common/fs/zfs/sys/dsl_dir.h:
sys/cddl/contrib/opensolaris/uts/common/fs/zfs/sys/zap_impl.h:
sys/cddl/contrib/opensolaris/uts/common/fs/zfs/sys/zap_leaf.h:
Remove now unused "phys pointer" aliases to db->db_data
from clients of the DMU buf user API.
Illumos issue:
5314 Remove "dbuf phys" db->db_data pointer aliases in ZFS
MFC after: 2 weeks
ZFS large block support.
Please note that booting from datasets that have recordsize greater
than 128KB is not supported (but it's Okay to enable the feature on
the pool). This *may* remain unchanged because of memory constraint.
Limited safety belt is provided for mounted root filesystem but use
caution is advised.
Illumos issue:
5027 zfs large block support
MFC after: 1 month
Encode CPU's number by XOR'ing the CPU ID against the 64-bit cpu_ticks().
Reviewed by: mav, gibbs
Differential Revision: https://phabric.freebsd.org/D521
MFC after: 2 weeks
Change dn->dn_dbufs from linked list to AVL tree.
Illumos issues:
4873 zvol unmap calls can take a very long time for larger datasets
MFC after: 2 weeks
Change the interaction between the DMU and ARC so that when the DMU is
shutting down an objset, we do not evict the data from the ARC. Instead
we simply coordinate the destruction of the DMU's data with the ARC.
The only case where we actually need to explicitly evict from the ARC is
when dbuf_rele_and_unlock() determines that the administrator has requested
that it not be kept in memory, via the primarycache/secondarycache properties.
In this case, we evict the data from the ARC by its blkptr_t, the same way
as when a block is freed we explicitly evict it from the ARC.
Illumos issue:
4631 zvol_get_stats triggering too many reads
MFC after: 2 weeks
4370 avoid transmitting holes during zfs send
4371 DMU code clean up
illumos/illumos-gate@43466aae47
NOTE: Make sure the boot code is updated if a zpool upgrade is
done on boot zpool.
MFC after: 2 weeks
4168 ztest assertion failure in dbuf_undirty
4169 verbatim import causes zdb to segfa
4170 zhack leaves pool in ACTIVE state
illumos/illumos-gate@7fdd916c47
MFC after: 2 weeks
illumos/illumos-gate@69962b5647
Please note the following changes:
- zio_ioctl has lost its priority parameter and now TRIM is executed
with 'now' priority
- some knobs are gone and some new knobs are added; not all of them are
exposed as tunables / sysctls yet
MFC after: 10 days
Sponsored by: HybridCluster [merge]
