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>
To ensure the arc is behaving properly we need greater visibility
in to exactly how it's managing the systems memory. This patch
takes one step in that direction be adding the current arc_state_t
for the anon, mru, mru_ghost, mfu, and mfs_ghost lists. The l2
arc_state_t is already well represented in the arcstats.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Historically the internal zfs debug infrastructure has been
scattered throughout the code. Since we expect to start making
more use of this code this patch performs some cleanup.
* Consolidate the zfs debug infrastructure in the zfs_debug.[ch]
files. This includes moving the zfs_flags and zfs_recover
variables, plus moving the zfs_panic_recover() function.
* Remove the existing unused functionality in zfs_debug.c and
replace it with code which correctly utilized the spl logging
infrastructure.
* Remove the __dprintf() function from zfs_ioctl.c. This is
dead code, the dprintf() functionality in the kernel relies
on the spl log support.
* Remove dprintf() from hdr_recl(). This wasn't particularly
useful and was missing the required format specifier anyway.
* Subsequent patches should unify the dprintf() and zfs_dbgmsg()
functions.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
The Linux 3.1 kernel has introduced the concept of per-filesystem
shrinkers which are directly assoicated with a super block. Prior
to this change there was one shared global shrinker.
The zfs code relied on being able to call the global shrinker when
the arc_meta_limit was exceeded. This would cause the VFS to drop
references on a fraction of the dentries in the dcache. The ARC
could then safely reclaim the memory used by these entries and
honor the arc_meta_limit. Unfortunately, when per-filesystem
shrinkers were added the old interfaces were made unavailable.
This change adds support to use the new per-filesystem shrinker
interface so we can continue to honor the arc_meta_limit. The
major benefit of the new interface is that we can now target
only the zfs filesystem for dentry and inode pruning. Thus we
can minimize any impact on the caching of other filesystems.
In the context of making this change several other important
issues related to managing the ARC were addressed, they include:
* The dnlc_reduce_cache() function which was called by the ARC
to drop dentries for the Posix layer was replaced with a generic
zfs_prune_t callback. The ZPL layer now registers a callback to
drop these dentries removing a layering violation which dates
back to the Solaris code. This callback can also be used by
other ARC consumers such as Lustre.
arc_add_prune_callback()
arc_remove_prune_callback()
* The arc_reduce_dnlc_percent module option has been changed to
arc_meta_prune for clarity. The dnlc functions are specific to
Solaris's VFS and have already been largely eliminated already.
The replacement tunable now represents the number of bytes the
prune callback will request when invoked.
* Less aggressively invoke the prune callback. We used to call
this whenever we exceeded the arc_meta_limit however that's not
strictly correct since it results in over zeleous reclaim of
dentries and inodes. It is now only called once the arc_meta_limit
is exceeded and every effort has been made to evict other data from
the ARC cache.
* More promptly manage exceeding the arc_meta_limit. When reading
meta data in to the cache if a buffer was unable to be recycled
notify the arc_reclaim thread to invoke the required prune.
* Added arcstat_prune kstat which is incremented when the ARC
is forced to request that a consumer prune its cache. Remember
this will only occur when the ARC has no other choice. If it
can evict buffers safely without invoking the prune callback
it will.
* This change is also expected to resolve the unexpect collapses
of the ARC cache. This would occur because when exceeded just the
arc_meta_limit reclaim presure would be excerted on the arc_c
value via arc_shrink(). This effectively shrunk the entire cache
when really we just needed to reclaim meta data.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#466Closes#292
In the upstream OpenSolaris ZFS code the maximum ARC usage is
limited to 3/4 of memory or all but 1GB, whichever is larger.
Because of how Linux's VM subsystem is organized these defaults
have proven to be too large which can lead to stability issues.
To avoid making everyone manually tune the ARC the defaults are
being changed to 1/2 of memory or all but 4GB. The rational for
this is as follows:
* Desktop Systems (less than 8GB of memory)
Limiting the ARC to 1/2 of memory is desirable for desktop
systems which have highly dynamic memory requirements. For
example, launching your web browser can suddenly result in a
demand for several gigabytes of memory. This memory must be
reclaimed from the ARC cache which can take some time. The
user will experience this reclaim time as a sluggish system
with poor interactive performance. Thus in this case it is
preferable to leave the memory as free and available for
immediate use.
* Server Systems (more than 8GB of memory)
Using all but 4GB of memory for the ARC is preferable for
server systems. These systems often run with minimal user
interaction and have long running daemons with relatively
stable memory demands. These systems will benefit most by
having as much data cached in memory as possible.
These values should work well for most configurations. However,
if you have a desktop system with more than 8GB of memory you may
wish to further restrict the ARC. This can still be accomplished
by setting the 'zfs_arc_max' module option.
Additionally, keep in mind these aren't currently hard limits.
The ARC is based on a slab implementation which can suffer from
memory fragmentation. Because this fragmentation is not visible
from the ARC it may believe it is within the specified limits while
actually consuming slightly more memory. How much more memory get's
consumed will be determined by how badly fragmented the slabs are.
In the long term this can be mitigated by slab defragmentation code
which was OpenSolaris solution. Or preferably, using the page cache
to back the ARC under Linux would be even better. See issue #75
for the benefits of more tightly integrating with the page cache.
This change also fixes a issue where the default ARC max was being
set incorrectly for machines with less than 2GB of memory. The
constant in the arc_c_max comparison must be explicitly cast to
a uint64_t type to prevent overflow and the wrong conditional
branch being taken. This failure was typically observed in VMs
which are commonly created with less than 2GB of memory.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Issue #75
The performance of the L2ARC can be tweaked by a number of tunables, which
may be necessary for different workloads:
l2arc_write_max max write bytes per interval
l2arc_write_boost extra write bytes during device warmup
l2arc_noprefetch skip caching prefetched buffers
l2arc_headroom number of max device writes to precache
l2arc_feed_secs seconds between L2ARC writing
l2arc_feed_min_ms min feed interval in milliseconds
l2arc_feed_again turbo L2ARC warmup
l2arc_norw no reads during writes
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#316
To accomindate the updated Linux 3.0 shrinker API the spl
shrinker compatibility code was updated. Unfortunately, this
couldn't be done cleanly without slightly adjusting the comapt
API. See spl commit a55bcaad18.
This commit updates the ZFS code to use the slightly modified
API. You must use the latest SPL if your building ZFS.
Normally when the arc_shrinker_func() function is called the return
value should be:
>=0 - To indicate the number of freeable objects in the cache, or
-1 - To indicate this cache should be skipped
However, when the shrinker callback is called with 'nr_to_scan' equal
to zero. The caller simply wants the number of freeable objects in
the cache and we must never return -1. This patch reorders the
first two conditionals in arc_shrinker_func() to ensure this behavior.
This patch also now explictly casts arc_size and arc_c_min to signed
int64_t types so MAX(x, 0) works as expected. As unsigned types
we would never see an negative value which defeated the purpose of
the MAX() lower bound and broke the shrinker logic.
Finally, when nr_to_scan is non-zero we explictly prevent all reclaim
below arc_c_min. This is done to prevent the Linux page cache from
completely crowding out the ARC. This limit is tunable and some
experimentation is likely going to be required to set it exactly right.
For now we're sticking with the OpenSolaris defaults.
Closes#218Closes#243
This commit adds module options for all existing zfs tunables.
Ideally the average user should never need to modify any of these
values. However, in practice sometimes you do need to tweak these
values for one reason or another. In those cases it's nice not to
have to resort to rebuilding from source. All tunables are visable
to modinfo and the list is as follows:
$ modinfo module/zfs/zfs.ko
filename: module/zfs/zfs.ko
license: CDDL
author: Sun Microsystems/Oracle, Lawrence Livermore National Laboratory
description: ZFS
srcversion: 8EAB1D71DACE05B5AA61567
depends: spl,znvpair,zcommon,zunicode,zavl
vermagic: 2.6.32-131.0.5.el6.x86_64 SMP mod_unload modversions
parm: zvol_major:Major number for zvol device (uint)
parm: zvol_threads:Number of threads for zvol device (uint)
parm: zio_injection_enabled:Enable fault injection (int)
parm: zio_bulk_flags:Additional flags to pass to bulk buffers (int)
parm: zio_delay_max:Max zio millisec delay before posting event (int)
parm: zio_requeue_io_start_cut_in_line:Prioritize requeued I/O (bool)
parm: zil_replay_disable:Disable intent logging replay (int)
parm: zfs_nocacheflush:Disable cache flushes (bool)
parm: zfs_read_chunk_size:Bytes to read per chunk (long)
parm: zfs_vdev_max_pending:Max pending per-vdev I/Os (int)
parm: zfs_vdev_min_pending:Min pending per-vdev I/Os (int)
parm: zfs_vdev_aggregation_limit:Max vdev I/O aggregation size (int)
parm: zfs_vdev_time_shift:Deadline time shift for vdev I/O (int)
parm: zfs_vdev_ramp_rate:Exponential I/O issue ramp-up rate (int)
parm: zfs_vdev_read_gap_limit:Aggregate read I/O over gap (int)
parm: zfs_vdev_write_gap_limit:Aggregate write I/O over gap (int)
parm: zfs_vdev_scheduler:I/O scheduler (charp)
parm: zfs_vdev_cache_max:Inflate reads small than max (int)
parm: zfs_vdev_cache_size:Total size of the per-disk cache (int)
parm: zfs_vdev_cache_bshift:Shift size to inflate reads too (int)
parm: zfs_scrub_limit:Max scrub/resilver I/O per leaf vdev (int)
parm: zfs_recover:Set to attempt to recover from fatal errors (int)
parm: spa_config_path:SPA config file (/etc/zfs/zpool.cache) (charp)
parm: zfs_zevent_len_max:Max event queue length (int)
parm: zfs_zevent_cols:Max event column width (int)
parm: zfs_zevent_console:Log events to the console (int)
parm: zfs_top_maxinflight:Max I/Os per top-level (int)
parm: zfs_resilver_delay:Number of ticks to delay resilver (int)
parm: zfs_scrub_delay:Number of ticks to delay scrub (int)
parm: zfs_scan_idle:Idle window in clock ticks (int)
parm: zfs_scan_min_time_ms:Min millisecs to scrub per txg (int)
parm: zfs_free_min_time_ms:Min millisecs to free per txg (int)
parm: zfs_resilver_min_time_ms:Min millisecs to resilver per txg (int)
parm: zfs_no_scrub_io:Set to disable scrub I/O (bool)
parm: zfs_no_scrub_prefetch:Set to disable scrub prefetching (bool)
parm: zfs_txg_timeout:Max seconds worth of delta per txg (int)
parm: zfs_no_write_throttle:Disable write throttling (int)
parm: zfs_write_limit_shift:log2(fraction of memory) per txg (int)
parm: zfs_txg_synctime_ms:Target milliseconds between tgx sync (int)
parm: zfs_write_limit_min:Min tgx write limit (ulong)
parm: zfs_write_limit_max:Max tgx write limit (ulong)
parm: zfs_write_limit_inflated:Inflated tgx write limit (ulong)
parm: zfs_write_limit_override:Override tgx write limit (ulong)
parm: zfs_prefetch_disable:Disable all ZFS prefetching (int)
parm: zfetch_max_streams:Max number of streams per zfetch (uint)
parm: zfetch_min_sec_reap:Min time before stream reclaim (uint)
parm: zfetch_block_cap:Max number of blocks to fetch at a time (uint)
parm: zfetch_array_rd_sz:Number of bytes in a array_read (ulong)
parm: zfs_pd_blks_max:Max number of blocks to prefetch (int)
parm: zfs_dedup_prefetch:Enable prefetching dedup-ed blks (int)
parm: zfs_arc_min:Min arc size (ulong)
parm: zfs_arc_max:Max arc size (ulong)
parm: zfs_arc_meta_limit:Meta limit for arc size (ulong)
parm: zfs_arc_reduce_dnlc_percent:Meta reclaim percentage (int)
parm: zfs_arc_grow_retry:Seconds before growing arc size (int)
parm: zfs_arc_shrink_shift:log2(fraction of arc to reclaim) (int)
parm: zfs_arc_p_min_shift:arc_c shift to calc min/max arc_p (int)
This change ensures the ARC meta-data limits are enforced. Without
this enforcement meta-data can grow to consume all of the ARC cache
pushing out data and hurting performance. The cache is aggressively
reclaimed but this is a soft and not a hard limit. The cache may
exceed the set limit briefly before being brought under control.
By default 25% of the ARC capacity can be used for meta-data. This
limit can be tuned by setting the 'zfs_arc_meta_limit' module option.
Once this limit is exceeded meta-data reclaim will occur in 3 percent
chunks, or may be tuned using 'arc_reduce_dnlc_percent'.
Closes#193
Under OpenSolaris all memory reclaim is done asyncronously. Under
Linux memory reclaim is done asynchronously _and_ synchronously.
When a process allocates memory with GFP_KERNEL it explicitly allows
the kernel to do reclaim on its behalf to satify the allocation.
If that GFP_KERNEL allocation fails the kernel may take more drastic
measures to reclaim the memory such as killing user space processes.
This was observed to happen with ZFS because the ARC could consume
a large fraction of the system memory but no synchronous reclaim
could be performed on it. The result was GFP_KERNEL allocations
could fail resulting in OOM events, and only moments latter the
arc_reclaim thread would free unused memory from the ARC.
This change leaves the arc_thread in place to manage the fundamental
ARC behavior. But it adds a synchronous (direct) reclaim path for
the ARC which can be called when memory is badly needed. It also
adds an asynchronous (indirect) reclaim path which is called
much more frequently to prune the ARC slab caches.
The following useful values were missing the arcstats. This change
adds them in to provide greater visibility in to the arcs behavior.
arc_no_grow 4 0
arc_tempreserve 4 0
arc_loaned_bytes 4 0
arc_meta_used 4 624774592
arc_meta_limit 4 400785408
arc_meta_max 4 625594176
It used to be the case that all KM_SLEEP allocations were GFS_NOFS.
Unfortunately this often resulted in the kernel being unable to
reclaim the ARC, inode, and dentry caches in a timely manor.
The fix was to make KM_SLEEP a GFP_KERNEL allocation in the SPL.
However, this increases the posibility of deadlocking the system
on a zfs write thread. If a zfs write thread attempts to perform
an allocation it may trigger synchronous reclaim. This reclaim
may attempt to flush dirty data/inode to disk to free memory.
Unforunately, this write cannot finish because the write thread
which would handle it is holding the previous transaction open.
Deadlock.
To avoid this all allocations in the zfs write thread path must
use KM_PUSHPAGE which prohibits synchronous reclaim for that
thread. In this way forward progress in ensured. The risk
with this change is I missed updating an allocation for the
write threads leaving an increased posibility of deadlock. If
any deadlocks remain they will be unlikely but we'll have to
make sure they all get fixed.
The issue is that cv_timedwait() sleeps uninterruptibly to block signals
and avoid waking up early. Under Linux this counts against the load
average keeping it artificially high. This change allows the arc to
sleep interruptibly which mean it may be woken up early due to a signal.
Normally this means some extra care must be taken to handle a potential
signal. But for the arcs usage of cv_timedwait() there is no harm in
waking up before the timeout expires so no extra handling is required.
This commit fixes a sign extension bug affecting l2arc devices. Extremely
large offsets may be passed down to the low level block device driver on
reads, generating errors similar to
attempt to access beyond end of device
sdbi1: rw=14, want=36028797014862705, limit=125026959
The unwanted sign extension occurrs because the function arc_read_nolock()
stores the offset as a daddr_t, a 32-bit signed int type in the Linux kernel.
This offset is then passed to zio_read_phys() as a uint64_t argument, causing
sign extension for values of 0x80000000 or greater. To avoid this, we store
the offset in a uint64_t.
This change also changes a few daddr_t struct members to uint64_t in the libspl
headers to avoid similar bugs cropping up in the future. We also add an ASSERT
to __vdev_disk_physio() to check for invalid offsets.
Closes#66
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Remove all instances of list handling where the API is not used
and instead list data members are directly accessed. Doing this
sort of thing is bad for portability.
Additionally, ensure that list_link_init() is called on newly
created list nodes. This ensures the node is properly initialized
and does not rely on the assumption that zero'ing the list_node_t
via kmem_zalloc() is the same as proper initialization.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Fix non-c90 compliant code, for the most part these changes
simply deal with where a particular variable is declared.
Under c90 it must alway be done at the very start of a block.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
