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
Setting the 'dist_' prefix is the correct way to instruct Automake
to include these files in the distribution. The EXTRA_DIST variable
is reserved for files which are not covered by the automatic rules.
http://www.gnu.org/software/automake/manual/automake.html#Basics
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Setting the 'dist_' prefix is the correct way to instruct Automake
to include these files in the distribution. The EXTRA_DIST variable
is reserved for files which are not covered by the automatic rules.
http://www.gnu.org/software/automake/manual/automake.html#Basics
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
The vdev_id udev helper currently applies slot renumbering rules to
every channel (JBOD) in the system. This is too inflexible for systems
with non-homogeneous storage topologies. The "slot" keyword now takes
an optional third parameter which names a channel to which the mapping
will apply. If the third parameter is omitted then the rule applies to
all channels. The first-specified rule that can match a slot takes
precedence. Therefore a channel-specific rule for a given slot should
generally appear before a generic rule for the same slot number. In
this way a custom slot mapping can be applied to a particular channel
and a default mapping applied to the rest.
Signed-off-by: Ned Bass <bass6@llnl.gov>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#2056
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
This is a first draft of a spl-module-parameters(5) man page. I have
just extracted the parameter name and its description with modinfo,
then checked the source what type it is and its default value.
This will need more work, preferably someone that actually know these
values and what to use them for. Similar to zfsonlinux/zfs#1856, but
for the spl.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closeszfsonlinux/zfs#1856
This is a first draft of a zfs-module-parameters(5) man page. I have
just extracted the parameter name and its description with modinfo,
then checked the source what type it is and its default value.
This will need more work, preferably someone that actually know these
values and what to use them for.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#1856
The zpool-features(5) man page should reference the Linux zfs(8)
and zpool(8) man pages. The 1M convention isn't used on Linux.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#1184
The zpool-features(5) man page was accidentally omitted from the
build target when feature flags was merged. As a result it doesn't
get installed as part of 'make install' so none of the packages
include this man page.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#1262
3035 LZ4 compression support in ZFS and GRUB
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Reviewed by: Christopher Siden <christopher.siden@delphix.com>
Reviewed by: George Wilson <george.wilson@delphix.com>
Approved by: Christopher Siden <csiden@delphix.com>
References:
illumos/illumos-gate@a6f561b4aehttps://www.illumos.org/issues/3035http://wiki.illumos.org/display/illumos/LZ4+Compression+In+ZFS
This patch has been slightly modified from the upstream Illumos
version to be compatible with Linux. Due to the very limited
stack space in the kernel a lz4 workspace kmem cache is used.
Since we are using gcc we are also able to take advantage of the
gcc optimized __builtin_ctz functions.
Support for GRUB has been dropped from this patch. That code
is available but those changes will need to made to the upstream
GRUB package.
Lastly, several hunks of dead code were dropped for clarity. They
include the functions real_LZ4_uncompress(), LZ4_compressBound()
and the Visual Studio specific hunks wrapped in _MSC_VER.
Ported-by: Eric Dillmann <eric@jave.fr>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#1217
2762 zpool command should have better support for feature flags
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Reviewed by: George Wilson <george.wilson@delphix.com>
Approved by: Eric Schrock <Eric.Schrock@delphix.com>
References:
illumos/illumos-gate@57221772c3https://www.illumos.org/issues/2762
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>
Add a vdev_id feature to map device names based on already defined
udev device links. To increase the odds that vdev_id will run after
the rules it depends on, increase the vdev.rules rule number from 60
to 69. With this change, vdev_id now provides functionality analogous
to zpool_id and zpool_layout, paving the way to retire those tools.
A defined alias takes precedence over a topology-derived name, but the
two naming methods can otherwise coexist. For example, one might name
drives in a JBOD with the sas_direct topology while naming an internal
L2ARC device with an alias.
For example, the following lines in vdev_id.conf will result in the
creation of links /dev/disk/by-vdev/{d1,d2}, each pointing to the same
target as the device link specified in the third field.
# by-vdev
# name fully qualified or base name of device link
alias d1 /dev/disk/by-id/wwn-0x5000c5002de3b9ca
alias d2 wwn-0x5000c5002def789e
Also perform some minor vdev_id cleanup, such as removal of the unused
-s command line option.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#981
Remove all of the generated autotools products from the repository
and update the .gitignore files accordingly.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#718
Currently, zvols have a discard granularity set to 0, which suggests to
the upper layer that discard requests of arbirarily small size and
alignment can be made efficiently.
In practice however, ZFS does not handle unaligned discard requests
efficiently: indeed, it is unable to free a part of a block. It will
write zeros to the specified range instead, which is both useless and
inefficient (see dnode_free_range).
With this patch, zvol block devices expose volblocksize as their discard
granularity, so the upper layer is aware that it's not supposed to send
discard requests smaller than volblocksize.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#862
The end_writeback() function was changed by moving the call to
inode_sync_wait() earlier in to evict(). This effecitvely changes
the ordering of the sync but it does not impact the details of
the zfs implementation.
However, as part of this change end_writeback() was renamed to
clear_inode() to reflect the new semantics. This change does
impact us and clear_inode() now maps to end_writeback() for
kernels prior to 3.5.
Signed-off-by: Richard Yao <ryao@cs.stonybrook.edu>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#784
The vmtruncate_range() support has been removed from the kernel in
favor of using the fallocate method in the file_operations table.
Signed-off-by: Richard Yao <ryao@cs.stonybrook.edu>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Issue #784
The export_operations member ->encode_fh() has been updated to
take both the child and parent inodes. This interface used to
take the child dentry and a bool describing if the parent is needed.
NOTE: While updating this code I noticed that we do not currently
cleanly handle the case where we're passed a connectable parent.
This code should be audited to make sure we're doing the right thing.
Signed-off-by: Richard Yao <ryao@cs.stonybrook.edu>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Issue #784
Currently, zpool online -e (dynamic vdev expansion) doesn't work on
whole disks because we're invoking ioctl(BLKRRPART) from userspace
while ZFS still has a partition open on the disk, which results in
EBUSY.
This patch moves the BLKRRPART invocation from the zpool utility to the
module. Specifically, this is done just before opening the device in
vdev_disk_open() which is called inside vdev_reopen(). This requires
jumping through some hoops to get to the disk device from the partition
device, and to make sure we can still open the partition after the
BLKRRPART call.
Note that this new code path is triggered on dynamic vdev expansion
only; other actions, like creating a new pool, are unchanged and still
call BLKRRPART from userspace.
This change also depends on API changes which are available in 2.6.37
and latter kernels. The build system has been updated to detect this,
but there is no compatibility mode for older kernels. This means that
online expansion will NOT be available in older kernels. However, it
will still be possible to expand the vdev offline.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#808
torvalds/linux@adc0e91ab1 introduced
introduced d_make_root() as a replacement for d_alloc_root(). Further
commits appear to have removed d_alloc_root() from the Linux source
tree. This causes the following failure:
error: implicit declaration of function 'd_alloc_root'
[-Werror=implicit-function-declaration]
To correct this we update the code to use the current d_make_root()
interface for readability. Then we introduce an autotools check
to determine if d_make_root() is available. If it isn't then we
define some compatibility logic which used the older d_alloc_root()
interface.
Signed-off-by: Richard Yao <ryao@gentoo.org>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#776
vdev_id parses the file /etc/zfs/vdev_id.conf to map a physical path
in a storage topology to a channel name. The channel name is combined
with a disk enclosure slot number to create an alias that reflects the
physical location of the drive. This is particularly helpful when it
comes to tasks like replacing failed drives. Slot numbers may also be
re-mapped in case the default numbering is unsatisfactory. The drive
aliases will be created as symbolic links in /dev/disk/by-vdev.
The only currently supported topologies are sas_direct and sas_switch:
o sas_direct - a channel is uniquely identified by a PCI slot and a
HBA port
o sas_switch - a channel is uniquely identified by a SAS switch port
A multipath mode is supported in which dm-mpath devices are handled by
examining the first running component disk, as reported by 'multipath
-l'. In multipath mode the configuration file should contain a
channel definition with the same name for each path to a given
enclosure.
vdev_id can replace the existing zpool_id script on systems where the
storage topology conforms to sas_direct or sas_switch. The script
could be extended to support other topologies as well. The advantage
of vdev_id is that it is driven by a single static input file that can
be shared across multiple nodes having a common storage toplogy.
zpool_id, on the other hand, requires a unique /etc/zfs/zdev.conf per
node and a separate slot-mapping file. However, zpool_id provides the
flexibility of using any device names that show up in
/dev/disk/by-path, so it may still be needed on some systems.
vdev_id's functionality subsumes that of the sas_switch_id script, and
it is unlikely that anyone is using it, so sas_switch_id is removed.
Finally, /dev/disk/by-vdev is added to the list of directories that
'zpool import' will scan.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#713
