For specific workloads consisting mainly of mfu data and new anon data
buffers, the aggressive growth of arc_p found in the arc_get_data_buf()
function can have detrimental effects on the mfu list size and ghost
list hit rate.
Running a workload consisting of two processes:
* Process 1 is creating many small files
* Process 2 is tar'ing a directory consisting of many small files
I've seen arc_p and the mru grow to their maximum size, while the mru
ghost list receives 100K times fewer hits than the mfu ghost list.
Ideally, as the mfu ghost list receives hits, arc_p should be driven
down and the size of the mfu should increase. Given the specific
workload I was testing with, the mfu list size should grow to a point
where almost no mfu ghost list hits would occur. Unfortunately, this
does not happen because the newly dirtied anon buffers constancy drive
arc_p to its maximum value and keep it there (effectively prioritizing
the mru list and starving the mfu list down to a negligible size).
The logic to increment arc_p from within the arc_get_data_buf() function
was introduced many years ago in this upstream commit:
commit 641fbdae3a027d12b3c3dcd18927ccafae6d58bc
Author: maybee <none@none>
Date: Wed Dec 20 15:46:12 2006 -0800
6505658 target MRU size (arc.p) needs to be adjusted more aggressively
and since I don't fully understand the motivation for the change, I am
reluctant to completely remove it.
As a way to test out how it's removal might affect performance, I've
disabled that code by default, but left it tunable via a module option.
Thus, if its removal is found to be grossly detrimental for certain
workloads, it can be re-enabled on the fly, without a code change.
Signed-off-by: Prakash Surya <surya1@llnl.gov>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Issue #2110
In an attempt to prevent arc_c from collapsing "too fast", the
arc_shrink() function was updated to take a "bytes" parameter by this
change:
commit 302f753f16
Author: Brian Behlendorf <behlendorf1@llnl.gov>
Date: Tue Mar 13 14:29:16 2012 -0700
Integrate ARC more tightly with Linux
Unfortunately, that change failed to make a similar change to the way
that arc_p was updated. So, there still exists the possibility for arc_p
to collapse to near 0 when the kernel start calling the arc's shrinkers.
This change attempts to fix this, by decrementing arc_p by the "bytes"
parameter in the same way that arc_c is updated.
In addition, a minimum value of arc_p is attempted to be maintained,
similar to the way a minimum arc_p value is maintained in arc_adapt().
Signed-off-by: Prakash Surya <surya1@llnl.gov>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Issue #2110
Decrease the mimimum ARC size from 1/32 of total system memory
(or 64MB) to a much smaller 4MB.
1) Large systems with over a 1TB of memory are being deployed
and reserving 1/32 of this memory (32GB) as the mimimum
requirement is overkill.
2) Tiny systems like the raspberry pi may only have 256MB of
memory in which case 64MB is far too large.
The ARC should be reclaimable if the VFS determines it needs
the memory for some other purpose. If you want to ensure the
ARC is never completely reclaimed due to memory pressure you
may still set a larger value with zfs_arc_min.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Prakash Surya <surya1@llnl.gov>
Issue #2110
Commit e0b0ca9 accidentally corrupted the l2_asize displayed in
arcstats. This was caused by changing the l2arc_buf_hdr.b_asize
member from an int to uint32_t type. There are places in the
code where this field is cast to a uint64_t resulting in the
b_hits member being treated as part of b_asize.
To resolve the issue the type has been changed to a uint64_t,
and the b_hits member is placed after the enum to prevent the
size of the structure from increasing.
This is a good example of exactly why it's a bad idea to use
ambiguous types (int) in these structures.
Signed-off-by: DHE <git@dehacked.net>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#1990
Back the allocations for ddt tables+entries and l2arc headers with
kmem caches. This will reduce the cost of allocating these commonly
used structures and allow for greater visibility of them through the
/proc/spl/kmem/slab interface.
Signed-off-by: John Layman <jlayman@sagecloud.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#1893
The vast majority of these changes are in Linux specific code.
They are the result of not having an automated style checker to
validate the code when it was originally written. Others were
caused when the common code was slightly adjusted for Linux.
This patch contains no functional changes. It only refreshes
the code to conform to style guide.
Everyone submitting patches for inclusion upstream should now
run 'make checkstyle' and resolve any warning prior to opening
a pull request. The automated builders have been updated to
fail a build if when 'make checkstyle' detects an issue.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#1821
The MAX when initializing arc_c_max doesn't make any sense because
it hasn't been set anywhere before. Though, arc_c_max should be
implicitly set to zero when initializing arc_stats, so the MAX
doesn't make any difference.
The MAX was mistakenly left if place when the Illumos default
values were changed for Linux.
Signed-off-by: david.chen <tuxoko@gmail.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#1941
4045 zfs write throttle & i/o scheduler performance work
1. The ZFS i/o scheduler (vdev_queue.c) now divides i/os into 5 classes: sync
read, sync write, async read, async write, and scrub/resilver. The scheduler
issues a number of concurrent i/os from each class to the device. Once a class
has been selected, an i/o is selected from this class using either an elevator
algorithem (async, scrub classes) or FIFO (sync classes). The number of
concurrent async write i/os is tuned dynamically based on i/o load, to achieve
good sync i/o latency when there is not a high load of writes, and good write
throughput when there is. See the block comment in vdev_queue.c (reproduced
below) for more details.
2. The write throttle (dsl_pool_tempreserve_space() and
txg_constrain_throughput()) is rewritten to produce much more consistent delays
when under constant load. The new write throttle is based on the amount of
dirty data, rather than guesses about future performance of the system. When
there is a lot of dirty data, each transaction (e.g. write() syscall) will be
delayed by the same small amount. This eliminates the "brick wall of wait"
that the old write throttle could hit, causing all transactions to wait several
seconds until the next txg opens. One of the keys to the new write throttle is
decrementing the amount of dirty data as i/o completes, rather than at the end
of spa_sync(). Note that the write throttle is only applied once the i/o
scheduler is issuing the maximum number of outstanding async writes. See the
block comments in dsl_pool.c and above dmu_tx_delay() (reproduced below) for
more details.
This diff has several other effects, including:
* the commonly-tuned global variable zfs_vdev_max_pending has been removed;
use per-class zfs_vdev_*_max_active values or zfs_vdev_max_active instead.
* the size of each txg (meaning the amount of dirty data written, and thus the
time it takes to write out) is now controlled differently. There is no longer
an explicit time goal; the primary determinant is amount of dirty data.
Systems that are under light or medium load will now often see that a txg is
always syncing, but the impact to performance (e.g. read latency) is minimal.
Tune zfs_dirty_data_max and zfs_dirty_data_sync to control this.
* zio_taskq_batch_pct = 75 -- Only use 75% of all CPUs for compression,
checksum, etc. This improves latency by not allowing these CPU-intensive tasks
to consume all CPU (on machines with at least 4 CPU's; the percentage is
rounded up).
--matt
APPENDIX: problems with the current i/o scheduler
The current ZFS i/o scheduler (vdev_queue.c) is deadline based. The problem
with this is that if there are always i/os pending, then certain classes of
i/os can see very long delays.
For example, if there are always synchronous reads outstanding, then no async
writes will be serviced until they become "past due". One symptom of this
situation is that each pass of the txg sync takes at least several seconds
(typically 3 seconds).
If many i/os become "past due" (their deadline is in the past), then we must
service all of these overdue i/os before any new i/os. This happens when we
enqueue a batch of async writes for the txg sync, with deadlines 2.5 seconds in
the future. If we can't complete all the i/os in 2.5 seconds (e.g. because
there were always reads pending), then these i/os will become past due. Now we
must service all the "async" writes (which could be hundreds of megabytes)
before we service any reads, introducing considerable latency to synchronous
i/os (reads or ZIL writes).
Notes on porting to ZFS on Linux:
- zio_t gained new members io_physdone and io_phys_children. Because
object caches in the Linux port call the constructor only once at
allocation time, objects may contain residual data when retrieved
from the cache. Therefore zio_create() was updated to zero out the two
new fields.
- vdev_mirror_pending() relied on the depth of the per-vdev pending queue
(vq->vq_pending_tree) to select the least-busy leaf vdev to read from.
This tree has been replaced by vq->vq_active_tree which is now used
for the same purpose.
- vdev_queue_init() used the value of zfs_vdev_max_pending to determine
the number of vdev I/O buffers to pre-allocate. That global no longer
exists, so we instead use the sum of the *_max_active values for each of
the five I/O classes described above.
- The Illumos implementation of dmu_tx_delay() delays a transaction by
sleeping in condition variable embedded in the thread
(curthread->t_delay_cv). We do not have an equivalent CV to use in
Linux, so this change replaced the delay logic with a wrapper called
zfs_sleep_until(). This wrapper could be adopted upstream and in other
downstream ports to abstract away operating system-specific delay logic.
- These tunables are added as module parameters, and descriptions added
to the zfs-module-parameters.5 man page.
spa_asize_inflation
zfs_deadman_synctime_ms
zfs_vdev_max_active
zfs_vdev_async_write_active_min_dirty_percent
zfs_vdev_async_write_active_max_dirty_percent
zfs_vdev_async_read_max_active
zfs_vdev_async_read_min_active
zfs_vdev_async_write_max_active
zfs_vdev_async_write_min_active
zfs_vdev_scrub_max_active
zfs_vdev_scrub_min_active
zfs_vdev_sync_read_max_active
zfs_vdev_sync_read_min_active
zfs_vdev_sync_write_max_active
zfs_vdev_sync_write_min_active
zfs_dirty_data_max_percent
zfs_delay_min_dirty_percent
zfs_dirty_data_max_max_percent
zfs_dirty_data_max
zfs_dirty_data_max_max
zfs_dirty_data_sync
zfs_delay_scale
The latter four have type unsigned long, whereas they are uint64_t in
Illumos. This accommodates Linux's module_param() supported types, but
means they may overflow on 32-bit architectures.
The values zfs_dirty_data_max and zfs_dirty_data_max_max are the most
likely to overflow on 32-bit systems, since they express physical RAM
sizes in bytes. In fact, Illumos initializes zfs_dirty_data_max_max to
2^32 which does overflow. To resolve that, this port instead initializes
it in arc_init() to 25% of physical RAM, and adds the tunable
zfs_dirty_data_max_max_percent to override that percentage. While this
solution doesn't completely avoid the overflow issue, it should be a
reasonable default for most systems, and the minority of affected
systems can work around the issue by overriding the defaults.
- Fixed reversed logic in comment above zfs_delay_scale declaration.
- Clarified comments in vdev_queue.c regarding when per-queue minimums take
effect.
- Replaced dmu_tx_write_limit in the dmu_tx kstat file
with dmu_tx_dirty_delay and dmu_tx_dirty_over_max. The first counts
how many times a transaction has been delayed because the pool dirty
data has exceeded zfs_delay_min_dirty_percent. The latter counts how
many times the pool dirty data has exceeded zfs_dirty_data_max (which
we expect to never happen).
- The original patch would have regressed the bug fixed in
zfsonlinux/zfs@c418410, which prevented users from setting the
zfs_vdev_aggregation_limit tuning larger than SPA_MAXBLOCKSIZE.
A similar fix is added to vdev_queue_aggregate().
- In vdev_queue_io_to_issue(), dynamically allocate 'zio_t search' on the
heap instead of the stack. In Linux we can't afford such large
structures on the stack.
Reviewed by: George Wilson <george.wilson@delphix.com>
Reviewed by: Adam Leventhal <ahl@delphix.com>
Reviewed by: Christopher Siden <christopher.siden@delphix.com>
Reviewed by: Ned Bass <bass6@llnl.gov>
Reviewed by: Brendan Gregg <brendan.gregg@joyent.com>
Approved by: Robert Mustacchi <rm@joyent.com>
References:
http://www.illumos.org/issues/4045illumos/illumos-gate@69962b5647
Ported-by: Ned Bass <bass6@llnl.gov>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#1913
3995 Memory leak of compressed buffers in l2arc_write_done
References:
https://illumos.org/issues/3995
Ported-by: Richard Yao <ryao@gentoo.org>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#1688
Issue #1775
3112 ztest does not honor ZFS_DEBUG
3113 ztest should use watchpoints to protect frozen arc bufs
3114 some leaked nvlists in zfsdev_ioctl
Reviewed by: Adam Leventhal <ahl@delphix.com>
Reviewed by: Matt Amdur <Matt.Amdur@delphix.com>
Reviewed by: George Wilson <george.wilson@delphix.com>
Reviewed by: Christopher Siden <chris.siden@delphix.com>
Approved by: Eric Schrock <eric.schrock@delphix.com>
References:
https://www.illumos.org/issues/3112https://www.illumos.org/issues/3113https://www.illumos.org/issues/3114illumos/illumos-gate@cd1c8b85eb
The /proc/self/cmd watchpoint interface is specific to Solaris.
Therefore, the #3113 implementation was reworked to use the more
portable mprotect(2) system call. When the pages are watched they
are marked read-only for protection. Any write to the protected
address range immediately trigger a SIGSEGV. The pages are marked
writable again when they are unwatched.
Ported-by: Brian Behlendorf <behlendorf1@llnl.gov>
Issue #1489
3236 zio nop-write
Reviewed by: Matt Ahrens <matthew.ahrens@delphix.com>
Reviewed by: Adam Leventhal <ahl@delphix.com>
Reviewed by: Christopher Siden <chris.siden@delphix.com>
Approved by: Garrett D'Amore <garrett@damore.org>
References:
illumos/illumos-gate@80901aea8ehttps://www.illumos.org/issues/3236
Porting Notes
1. This patch is being merged dispite an increased instance of
https://www.illumos.org/issues/3113 being triggered by ztest.
Ported-by: Brian Behlendorf <behlendorf1@llnl.gov>
Issue #1489
3742 zfs comments need cleaner, more consistent style
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Reviewed by: George Wilson <george.wilson@delphix.com>
Reviewed by: Eric Schrock <eric.schrock@delphix.com>
Approved by: Christopher Siden <christopher.siden@delphix.com>
References:
https://www.illumos.org/issues/3742illumos/illumos-gate@f717074149
Ported-by: Richard Yao <ryao@gentoo.org>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Issue #1775
Porting notes:
1. The change to zfs_vfsops.c was dropped because it involves
zfs_mount_label_policy, which does not exist in the Linux port.
3741 zfs needs better comments
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Reviewed by: Eric Schrock <eric.schrock@delphix.com>
Approved by: Christopher Siden <christopher.siden@delphix.com>
References:
https://www.illumos.org/issues/3741illumos/illumos-gate@3e30c24aee
Ported-by: Richard Yao <ryao@gentoo.org>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Issue #1775
3598 want to dtrace when errors are generated in zfs
Reviewed by: Dan Kimmel <dan.kimmel@delphix.com>
Reviewed by: Adam Leventhal <ahl@delphix.com>
Reviewed by: Christopher Siden <christopher.siden@delphix.com>
Approved by: Garrett D'Amore <garrett@damore.org>
References:
https://www.illumos.org/issues/3598illumos/illumos-gate@be6fd75a69
Ported-by: Richard Yao <ryao@gentoo.org>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Issue #1775
Porting notes:
1. include/sys/zfs_context.h has been modified to render some new
macros inert until dtrace is available on Linux.
2. Linux-specific changes have been adapted to use SET_ERROR().
3. I'm NOT happy about this change. It does nothing but ugly
up the code under Linux. Unfortunately we need to take it to
avoid more merge conflicts in the future. -Brian
3561 arc_meta_limit should be exposed via kstats
3116 zpool reguid may log negative guids to internal SPA history
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Reviewed by: Adam Leventhal <ahl@delphix.com>
Reviewed by: Christopher Siden <christopher.siden@delphix.com>
Reviewed by: Gordon Ross <gordon.ross@nexenta.com>
Approved by: Garrett D'Amore <garrett@damore.org>
References:
https://www.illumos.org/issues/3561https://www.illumos.org/issues/3116illumos/illumos-gate@20128a0826
Ported-by: Richard Yao <ryao@gentoo.org>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Porting Notes:
1. The spa change was accidentally included in the libzfs_core merge.
2. "Add missing arcstats" (1834f2d8b7)
already implemented these kstats a few years ago.
3522 zfs module should not allow uninitialized variables
Reviewed by: Sebastien Roy <seb@delphix.com>
Reviewed by: Adam Leventhal <ahl@delphix.com>
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Approved by: Garrett D'Amore <garrett@damore.org>
References:
https://www.illumos.org/issues/3522illumos/illumos-gate@d5285cae91
Ported-by: Richard Yao <ryao@gentoo.org>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Porting notes:
1. ZFSOnLinux had already addressed many of these issues because of
its use of -Wall. However, the manner in which they were addressed
differed. The illumos fixes replace the ones previously made in
ZFSOnLinux to reduce code differences.
2. Part of the upstream patch made a small change to arc.c that might
address zfsonlinux/zfs#1334.
3. The initialization of aclsize in zfs_log_create() differs because
vsecp is a NULL pointer on ZFSOnLinux.
4. The changes to zfs_register_callbacks() were dropped because it
has diverged and needs to be resynced.
Currently there is no mechanism to inspect which dbufs are being
cached by the system. There are some coarse counters in arcstats
by they only give a rough idea of what's being cached. This patch
aims to improve the current situation by adding a new dbufs kstat.
When read this new kstat will walk all cached dbufs linked in to
the dbuf_hash. For each dbuf it will dump detailed information
about the buffer. It will also dump additional information about
the referenced arc buffer and its related dnode. This provides a
more complete view in to exactly what is being cached.
With this generic infrastructure in place utilities can be written
to post-process the data to understand exactly how the caching is
working. For example, the data could be processed to show a list
of all cached dnodes and how much space they're consuming. Or a
similar list could be generated based on dnode type. Many other
ways to interpret the data exist based on what kinds of questions
you're trying to answer.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Prakash Surya <surya1@llnl.gov>
This change is an attempt to add visibility into the arc_read calls
occurring on a system, in real time. To do this, a list was added to the
in memory SPA data structure for a pool, with each element on the list
corresponding to a call to arc_read. These entries are then exported
through the kstat interface, which can then be interpreted in userspace.
For each arc_read call, the following information is exported:
* A unique identifier (uint64_t)
* The time the entry was added to the list (hrtime_t)
(*not* wall clock time; relative to the other entries on the list)
* The objset ID (uint64_t)
* The object number (uint64_t)
* The indirection level (uint64_t)
* The block ID (uint64_t)
* The name of the function originating the arc_read call (char[24])
* The arc_flags from the arc_read call (uint32_t)
* The PID of the reading thread (pid_t)
* The command or name of thread originating read (char[16])
From this exported information one can see, in real time, exactly what
is being read, what function is generating the read, and whether or not
the read was found to be already cached.
There is still some work to be done, but this should serve as a good
starting point.
Specifically, dbuf_read's are not accounted for in the currently
exported information. Thus, a follow up patch should probably be added
to export these calls that never call into arc_read (they only hit the
dbuf hash table). In addition, it might be nice to create a utility
similar to "arcstat.py" to digest the exported information and display
it in a more readable format. Or perhaps, log the information and allow
for it to be "replayed" at a later time.
Signed-off-by: Prakash Surya <surya1@llnl.gov>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
This reverts commit fadd0c4da1 which
introduced a regression in honoring the meta limit.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Close#1660
When the meta limit is exceeded the ARC evicts some meta data
buffers from the mfu+mru lists. Unfortunately, for meta data
heavy workloads it's possible for these buffers to accumulate
on the ghost lists if arc_c doesn't exceed arc_size.
To handle this case arc_adjust_meta() has been entended to
explicitly evict meta data buffers from the ghost lists in
proportion to what was evicted from the mfu+mru lists.
If this is insufficient we request that the VFS release
some inodes and dentries. This will result in the release
of some dnodes which are counted as 'other' metadata.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
The default behavior of arc_evict_ghost() is to start by evicting
data buffers. Then only if the requested number of bytes to evict
cannot be satisfied by data buffers move on to meta data buffers.
This is ideal for honoring arc_c since it's preferable to keep the
meta data cached. However, if we're trying to free memory from the
arc to honor the meta limit it's a problem because we will need to
discard all the data to get to the meta data.
To avoid this issue the arc_evict_ghost() is now passed a fourth
argumented describing which buffer type to start with. The
arc_evict() function already behaves exactly like this for a
same reason so this is consistent with the existing code.
All existing callers have been updated to pass ARC_BUFC_DATA so
this patch introduces no functional change. New callers may
pass ARC_BUFC_METADATA to skip immediately to evicting meta
data leaving the normal data untouched.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
3137 L2ARC compression
Reviewed by: George Wilson <george.wilson@delphix.com>
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Approved by: Dan McDonald <danmcd@nexenta.com>
References:
illumos/illumos-gate@aad02571bchttps://www.illumos.org/issues/3137http://wiki.illumos.org/display/illumos/L2ARC+Compression
Notes for Linux port:
A l2arc_nocompress module option was added to prevent the
compression of l2arc buffers regardless of how a dataset's
compression property is set. This allows the legacy behavior
to be preserved.
Ported by: James H <james@kagisoft.co.uk>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#1379
This is analogous to SPL commit zfsonlinux/spl@b9b3715. While
we don't have clear evidence of systems getting caught here
indefinately like in the SPL this ensures that it will never
happen.
Signed-off-by: Richard Yao <ryao@gentoo.org>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#1579
The code involving b_thawed appears to be dead, so lets discard it.
Signed-off-by: Richard Yao <ryao@gentoo.org>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Issue #1614
These functions are used in neither Illumos nor ZFSOnLinux. They appear
to have been replaced by arc_buf_alloc()/arc_buf_free(), so lets remove
them.
Signed-off-by: Richard Yao <ryao@gentoo.org>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Issue #1614
The l2arc module options can be made safely writable. This allows
the options to be changed without unloading/loading the modules.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
These days modern SSDs can efficiently service concurrent reads
and writes. When this flag was added that wasn't really the
case for a variety of SSD controllers. But now we can set the
default value to take advantage of this parallelism and only
disable this as needed for specific troublesome hardware.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Based on the comments in arc.c we know that buffers can exist both
in arc and l2arc, under this circumstance both arc_buf_hdr_t and
l2arc_buf_hdr_t will be allocated. However the current logic only
cares for memory that l2arc_buf_hdr takes up when the buffer's
state transfers from or to arc_l2c_only. This will cause obvious
deviations for illumos's zfs version since the sizeof(l2arc_buf_hdr)
is larger than ZOL's. We can implement the calcuation in the
following simple way:
1. When allocate a l2arc_buf_hdr_t we add its memory consumption
instantly and subtract it when we free or evict the l2arc buf.
2. According to l2arc_hdr_stat_add and l2arc_hdr_stat_remove, if
the buffer only stays in l2arc we should also add the memory
its arc_buf_hdr_t consumes, so we only need to add HDR_SIZE to
arcstat_l2_hdr_size since we already concerned with L2HDR_SIZE
in step 1 and the same for transfering arc bufs from l2arc only
state.
The testbox has 2 4-core Intel Xeon CPUs(2.13GHz), with 16GB memory
and tests were set upped in the following way:
1. Fdisked a SATA disk into two partitions, one partition for zpool
storage and the other one was used as the cache device.
2. Generated some files occupying 14GB altogether in the zpool
prepared in step 1 using iozone.
3. Read them all using md5sum and watched the l2arc related statistics
in /proc/spl/kstat/zfs/arcstats. After the reading ended the
l2_hdr_size and l2_size were shown like this:
l2_size 4 4403780608
l2_hdr_size 4 0
which was weird.
4. After applying this patch and reran step 1-3, the results were
as following:
l2_size 4 4306443264
l2_hdr_size 4 535600
these numbers made sense, on 64-bit systems the
sizeof(l2arc_buf_hdr_t) is 16 bytes. Assue all blocks cached by
l2arc are 128KB, so 535600/16*128*1024=4387635200, since not all
blocks are equal-sized, the theoretical result will be a little
bigger, as we can see.
Since I'm familiar with systemtap instrumentation tool I used it to
examine what had happened. The script looked like this:
probe module("zfs").function("arc_chage_state")
{
if ($new_state == $arc_l2_only)
printf("change arc buf to arc_l2_only\n")
}
It will print out some information each time we call funciton
arc_chage_state if the argument new_state is arc_l2_only. I
gathered the trace logs and found that none of the arc bufs ran
into arc state arc_l2_only when the tests was running, this was
the reason why l2_hdr_size in step 3 was 0. The arc bufs fell into
arc_l2_only when the pool or the filesystem was offlined.
Signed-off-by: Ying Zhu <casualfisher@gmail.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
When we remove references of arc bufs in the arc_anon state we
needn't take its header's hash_lock, so postpone it to where we
really need it to avoid unnecessary invocations of function buf_hash.
Signed-off-by: Ying Zhu <casualfisher@gmail.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#1557
3805 arc shouldn't cache freed blocks
Reviewed by: George Wilson <george.wilson@delphix.com>
Reviewed by: Christopher Siden <christopher.siden@delphix.com>
Reviewed by: Richard Elling <richard.elling@dey-sys.com>
Reviewed by: Will Andrews <will@firepipe.net>
Approved by: Dan McDonald <danmcd@nexenta.com>
References:
illumos/illumos-gate@6e6d5868f5https://www.illumos.org/issues/3805
ZFS should proactively evict freed blocks from the cache.
On dcenter, we saw that we were caching ~256GB of metadata, while the
pool only had <4GB of metadata on disk. We were wasting about half the
system's RAM (252GB) on blocks that have been freed.
Even though these freed blocks will never be used again, and thus will
eventually be evicted, this causes us to use memory inefficiently for 2
reasons:
1. A block that is freed has no chance of being accessed again, but will
be kept in memory preferentially to a block that was accessed before it
(and is thus older) but has not been freed and thus has at least some
chance of being accessed again.
2. We partition the ARC into several buckets:
user data that has been accessed only once (MRU)
metadata that has been accessed only once (MRU)
user data that has been accessed more than once (MFU)
metadata that has been accessed more than once (MFU)
The user data vs metadata split is somewhat arbitrary, and the primary
control on how much memory is used to cache data vs metadata is to
simply try to keep the proportion the same as it has been in the past
(each bucket "evicts against" itself). The secondary control is to
evict data before evicting metadata.
Because of this bucketing, we may end up with one bucket mostly
containing freed blocks that are very old, while another bucket has more
recently accessed, still-allocated blocks. Data in the useful bucket
(with still-allocated blocks) may be evicted in preference to data in
the useless bucket (with old, freed blocks).
On dcenter, we saw that the MFU metadata bucket was 230MB, while the MFU
data bucket was 27GB and the MRU metadata bucket was 256GB. However,
the vast majority of data in the MRU metadata bucket (256GB) was freed
blocks, and thus useless. Meanwhile, the MFU metadata bucket (230MB)
was constantly evicting useful blocks that will be soon needed.
The problem of cache segmentation is a larger problem that needs more
investigation. However, if we stop caching freed blocks, it should
reduce the impact of this more fundamental issue.
Ported-by: Richard Yao <ryao@cs.stonybrook.edu>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#1503
3006 VERIFY[S,U,P] and ASSERT[S,U,P] frequently check if first
argument is zero
Reviewed by Matt Ahrens <matthew.ahrens@delphix.com>
Reviewed by George Wilson <george.wilson@delphix.com>
Approved by Eric Schrock <eric.schrock@delphix.com>
References:
illumos/illumos-gate@fb09f5aad4https://illumos.org/issues/3006
Requires:
zfsonlinux/spl@1c6d149feb
Ported-by: Tim Chase <tim@chase2k.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#1509
The PaX team modified the kernel's modpost to report writeable function
pointers as section mismatches because they are potential exploit
targets. We could ignore the warnings, but their presence can obscure
actual issues. Proper const correctness can also catch programming
mistakes.
Building the kernel modules against a PaX/GrSecurity patched Linux 3.4.2
kernel reports 133 section mismatches prior to this patch. This patch
eliminates 130 of them. The quantity of writeable function pointers
eliminated by constifying each structure is as follows:
vdev_opts_t 52
zil_replay_func_t 24
zio_compress_info_t 24
zio_checksum_info_t 9
space_map_ops_t 7
arc_byteswap_func_t 5
The remaining 3 writeable function pointers cannot be addressed by this
patch. 2 of them are in zpl_fs_type. The kernel's sget function requires
that this be non-const. The final writeable function pointer is created
by SPL_SHRINKER_DECLARE. The kernel's set_shrinker() and
remove_shrinker() functions also require that this be non-const.
Signed-off-by: Richard Yao <ryao@cs.stonybrook.edu>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#1300
The zfs_arc_memory_throttle_disable module option was introduced
by commit 0c5493d470 to resolve a
memory miscalculation which could result in the txg_sync thread
spinning.
When this was first introduced the default behavior was left
unchanged until enough real world usage confirmed there were no
unexpected issues. We've now reached that point. Linux's
direct reclaim is working as expected so we're enabling this
behavior by default.
This helps pave the way to retire the spl_kmem_availrmem()
functionality in the SPL layer. This was the only caller.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Issue #938
The way in which virtual box ab(uses) memory can throw off the
free memory calculation in arc_memory_throttle(). The result is
the txg_sync thread will effectively spin waiting for memory to
be released even though there's lots of memory on the system.
To handle this case I'm adding a zfs_arc_memory_throttle_disable
module option largely for virtual box users. Setting this option
disables free memory checks which allows the txg_sync thread to
make progress.
By default this option is disabled to preserve the current
behavior. However, because Linux supports direct memory reclaim
it's doubtful throttling due to perceived memory pressure is ever
a good idea. We should enable this option by default once we've
done enough real world testing to convince ourselve there aren't
any unexpected side effects.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#938
Commit 1eb5bfa introduced a new zfs_disable_dup_eviction tunable.
It should have been made available as a module option in the
original patch but was overlooked.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
2618 arc.c mistypes in the comments
Reviewed by: Jason King <jason.brian.king@gmail.com>
Reviewed by: Josef Sipek <jeffpc@josefsipek.net>
Approved by: Richard Lowe <richlowe@richlowe.net>
References:
illumos/illumos-gate@fc98fea58e
illumos changeset: 13721:5b51a16a186f
https://www.illumos.org/issues/2618
Ported-by: Brian Behlendorf <behlendorf1@llnl.gov>
2619 asynchronous destruction of ZFS file systems
2747 SPA versioning with zfs feature flags
Reviewed by: Matt Ahrens <mahrens@delphix.com>
Reviewed by: George Wilson <gwilson@delphix.com>
Reviewed by: Richard Lowe <richlowe@richlowe.net>
Reviewed by: Dan Kruchinin <dan.kruchinin@gmail.com>
Approved by: Eric Schrock <Eric.Schrock@delphix.com>
References:
illumos/illumos-gate@53089ab7c8illumos/illumos-gate@ad135b5d64
illumos changeset: 13700:2889e2596bd6
https://www.illumos.org/issues/2619https://www.illumos.org/issues/2747
NOTE: The grub specific changes were not ported. This change
must be made to the Linux grub packages.
Ported-by: Brian Behlendorf <behlendorf1@llnl.gov>
There are currently three vmem_size() consumers all of which are
part of the ARC implemention. However, since the expected behavior
of the Linux and Solaris virtual memory subsystems are so different
the behavior in each of these instances needs to be reevaluated.
* arc_evict_needed() - This is actually dead code. Arena support
was never added to the SPL and zio_arena is always NULL. This
support isn't needed so we simply remove this dead code.
* arc_memory_throttle() - On Solaris where virtual memory constitutes
almost all of the address space we can reasonably expect there to be
a fairly large amount free. However, on Linux by default we only
have about 100MB total and that's heavily used by the ARC. So the
expectation on Linux is that this will usually be a small value.
Therefore we remove the vmem_size() check for i386 systems because
the expectation is that it will be less than the zfs_write_limit_max.
* arc_init() - Here vmem_size() is used to initially size the ARC.
Since the ARC is currently backed by the virtual address space it
makes sense to use this as a limit on the ARC for 32-bit systems.
This code can be removed when the ARC is backed by the page cache.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#831
Commit c409e4647f introduced a
number of module parameters. This required several types to be
changed to accomidate the required module parameters Linux macros.
Unfortunately, arc.c contained its own extern definition of the
zfs_write_limit_max variable and its type was not updated to be
consistent with its dsl_pool.c counterpart. If the variable had
been properly marked extern in a common header, then gcc would
have generated a warning and this would not have slipped through.
The result of this was that the ARC unconditionally expected
zfs_write_limit_max to be 64-bit. Unfortunately, the largest size
integer module parameter that Linux supports is unsigned long, which
varies in size depending on the host system's native word size. The
effect was that on 32-bit systems, ARC incorrectly performed 64-bit
operations on a 32-bit value by reading the neighboring 32 bits as
the upper 32 bits of the 64-bit value.
We correct that by changing the extern declaration to use the unsigned
long type and move these extern definitions in to the common arc.h
header. This should make ARC correctly treat zfs_write_limit_max as a
32-bit value on 32-bit systems.
Reported-by: Jorgen Lundman <lundman@lundman.net>
Signed-off-by: Richard Yao <ryao@cs.stonybrook.edu>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#749
This warning indicates the incorrect use of KM_SLEEP in a call
path which must use KM_PUSHPAGE to avoid deadlocking in direct
reclaim. See commit b8d06fca08
for additional details.
SPL: Fixing allocation for task txg_sync (6093) which
used GFP flags 0x297bda7c with PF_NOFS set
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Issue #917
Differences between how paging is done on Solaris and Linux can cause
deadlocks if KM_SLEEP is used in any the following contexts.
* The txg_sync thread
* The zvol write/discard threads
* The zpl_putpage() VFS callback
This is because KM_SLEEP will allow for direct reclaim which may result
in the VM calling back in to the filesystem or block layer to write out
pages. If a lock is held over this operation the potential exists to
deadlock the system. To ensure forward progress all memory allocations
in these contexts must us KM_PUSHPAGE which disables performing any I/O
to accomplish the memory allocation.
Previously, this behavior was acheived by setting PF_MEMALLOC on the
thread. However, that resulted in unexpected side effects such as the
exhaustion of pages in ZONE_DMA. This approach touchs more of the zfs
code, but it is more consistent with the right way to handle these cases
under Linux.
This is patch lays the ground work for being able to safely revert the
following commits which used PF_MEMALLOC:
21ade34 Disable direct reclaim for z_wr_* threads
cfc9a5c Fix zpl_writepage() deadlock
eec8164 Fix ASSERTION(!dsl_pool_sync_context(tx->tx_pool))
Signed-off-by: Richard Yao <ryao@cs.stonybrook.edu>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Issue #726
Reviewed by: George Wilson <gwilson@zfsmail.com>
Reviewed by: Igor Kozhukhov <ikozhukhov@gmail.com>
Reviewed by: Alexander Eremin <alexander.eremin@nexenta.com>
Reviewed by: Alexander Stetsenko <ams@nexenta.com>
Approved by: Richard Lowe <richlowe@richlowe.net>
References:
https://www.illumos.org/issues/1748
This commit modifies the user to kernel space ioctl ABI. Extra
care should be taken when updating to ensure both the kernel
modules and utilities are updated. If only the user space
component is updated both the 'zpool events' command and the
'zpool reguid' command will not work until the kernel modules
are updated.
Ported by: Martin Matuska <martin@matuska.org>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#665
23bdb07d4e updated the ARC memory limits
to be 1/2 of memory or all but 4GB. Unfortunately, these values assume
zero internal fragmentation in the SLUB allocator, when in reality, the
internal fragmentation could be as high as 50%, effectively doubling
memory usage. This poses clear safety issues, because it permits the
size of ARC to exceed system memory.
This patch changes this so that the default value of arc_c_max is always
1/2 of system memory. This effectively limits the ARC to the memory that
the system has physically installed.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#660
Under Solaris the ARC was designed to stay one step ahead of the
VM subsystem. It would attempt to recognize low memory situtions
before they occured and evict data from the cache. It would also
make assessments about if there was enough free memory to perform
a specific operation.
This was all possible because Solaris exposes a fairly decent
view of the memory state of the system to other kernel threads.
Linux on the other hand does not make this information easily
available. To avoid extensive modifications to the ARC the SPL
attempts to provide these same interfaces. While this works it
is not ideal and problems can arise when the ARC and Linux have
different ideas about when your out of memory. This has manifested
itself in the past as a spinning arc_reclaim_thread.
This patch abandons the emulated Solaris interfaces in favor of
the prefered Linux interface. That means moving the bulk of the
memory reclaim logic out of the arc_reclaim_thread and in to the
evict driven shrinker callback. The Linux VM will call this
function when it needs memory. The ARC is then responsible for
attempting to free the requested amount of memory if possible.
Several interfaces have been modified to accomidate this approach,
however the basic user space implementation remains the same.
The following changes almost exclusively just apply to the kernel
implementation.
* Removed the hdr_recl() reclaim callback which is redundant
with the broader arc_shrinker_func().
* Reduced arc_grow_retry to 5 seconds from 60. This is now used
internally in the ARC with arc_no_grow to indicate that direct
reclaim was recently performed. This typically indicates a
rapid change in memory demands which the kswapd threads were
unable to keep ahead of. As long as direct reclaim is happening
once every 5 seconds arc growth will be paused to avoid further
contributing to the existing memory pressure. The more common
indirect reclaim paths will not set arc_no_grow.
* arc_shrink() has been extended to take the number of bytes by
which arc_c should be reduced. This allows for a more granual
reduction of the arc target. Since the kernel provides a
reclaim value to the arc_shrinker_func() this value is used
instead of 1<<arc_shrink_shift.
* arc_reclaim_needed() has been removed. It was used to determine
if the system was under memory pressure and relied extensively
on Solaris specific VM interfaces. In most case the new code
just checks arc_no_grow which indicates that within the last
arc_grow_retry seconds direct memory reclaim occurred.
* arc_memory_throttle() has been updated to always include the
amount of evictable memory (arc and page cache) in its free
space calculations. This space is largely available in most
call paths due to direct memory reclaim.
* The Solaris pageout code was also removed to avoid confusion.
It has always been disabled due to proc_pageout being defined
as NULL in the Linux port.
Signed-off-by: Prakash Surya <surya1@llnl.gov>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
There is potential for deadlock in the l2arc_feed thread if KM_PUSHPAGE
is not used for the allocations made in l2arc_write_buffers.
Specifically, if KM_PUSHPAGE is not used for these allocations, it is
possible for reclaim to be triggered which can cause the l2arc_feed
thread to deadlock itself on the ARC_mru mutex. An example of this is
demonstrated in the following backtrace of the l2arc_feed thread:
crash> bt 4123
PID: 4123 TASK: ffff88062f8c1500 CPU: 6 COMMAND: "l2arc_feed"
0 [ffff88062511d610] schedule at ffffffff814eeee0
1 [ffff88062511d6d8] __mutex_lock_slowpath at ffffffff814f057e
2 [ffff88062511d748] mutex_lock at ffffffff814f041b
3 [ffff88062511d768] arc_evict at ffffffffa05130ca [zfs]
4 [ffff88062511d858] arc_adjust at ffffffffa05139a9 [zfs]
5 [ffff88062511d878] arc_shrink at ffffffffa0513a95 [zfs]
6 [ffff88062511d898] arc_kmem_reap_now at ffffffffa0513be8 [zfs]
7 [ffff88062511d8c8] arc_shrinker_func at ffffffffa0513ccc [zfs]
8 [ffff88062511d8f8] shrink_slab at ffffffff8112a17a
9 [ffff88062511d958] do_try_to_free_pages at ffffffff8112bfdf
10 [ffff88062511d9e8] try_to_free_pages at ffffffff8112c3ed
11 [ffff88062511da98] __alloc_pages_nodemask at ffffffff8112431d
12 [ffff88062511dbb8] kmem_getpages at ffffffff8115e632
13 [ffff88062511dbe8] fallback_alloc at ffffffff8115f24a
14 [ffff88062511dc68] ____cache_alloc_node at ffffffff8115efc9
15 [ffff88062511dcc8] __kmalloc at ffffffff8115fbf9
16 [ffff88062511dd18] kmem_alloc_debug at ffffffffa047b8cb [spl]
17 [ffff88062511dda8] l2arc_feed_thread at ffffffffa0511e71 [zfs]
18 [ffff88062511dea8] thread_generic_wrapper at ffffffffa047d1a1 [spl]
19 [ffff88062511dee8] kthread at ffffffff81090a86
20 [ffff88062511df48] kernel_thread at ffffffff8100c14a
Signed-off-by: Prakash Surya <surya1@llnl.gov>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Due to a typo the mru ghost lists stats were accidentally being
exposed as the mfu ghost list stats. This was harmless but
confusing since memory usage could be over reported.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Keep counters for the various reasons that a thread may end up
in txg_wait_open() waiting on a new txg. This can be useful
when attempting to determine why a particular workload is
under performing.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
