| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
Pull md updates from Neil Brown:
"More updates that usual this time. A few have performance impacts
which hould mostly be positive, but RAID5 (in particular) can be very
work-load ensitive... We'll have to wait and see.
Highlights:
- "experimental" code for managing md/raid1 across a cluster using
DLM. Code is not ready for general use and triggers a WARNING if
used. However it is looking good and mostly done and having in
mainline will help co-ordinate development.
- RAID5/6 can now batch multiple (4K wide) stripe_heads so as to
handle a full (chunk wide) stripe as a single unit.
- RAID6 can now perform read-modify-write cycles which should help
performance on larger arrays: 6 or more devices.
- RAID5/6 stripe cache now grows and shrinks dynamically. The value
set is used as a minimum.
- Resync is now allowed to go a little faster than the 'mininum' when
there is competing IO. How much faster depends on the speed of the
devices, so the effective minimum should scale with device speed to
some extent"
* tag 'md/4.1' of git://neil.brown.name/md: (58 commits)
md/raid5: don't do chunk aligned read on degraded array.
md/raid5: allow the stripe_cache to grow and shrink.
md/raid5: change ->inactive_blocked to a bit-flag.
md/raid5: move max_nr_stripes management into grow_one_stripe and drop_one_stripe
md/raid5: pass gfp_t arg to grow_one_stripe()
md/raid5: introduce configuration option rmw_level
md/raid5: activate raid6 rmw feature
md/raid6 algorithms: xor_syndrome() for SSE2
md/raid6 algorithms: xor_syndrome() for generic int
md/raid6 algorithms: improve test program
md/raid6 algorithms: delta syndrome functions
raid5: handle expansion/resync case with stripe batching
raid5: handle io error of batch list
RAID5: batch adjacent full stripe write
raid5: track overwrite disk count
raid5: add a new flag to track if a stripe can be batched
raid5: use flex_array for scribble data
md raid0: access mddev->queue (request queue member) conditionally because it is not set when accessed from dm-raid
md: allow resync to go faster when there is competing IO.
md: remove 'go_faster' option from ->sync_request()
...
|
|
When array is degraded, read data landed on failed drives will result in
reading rest of data in a stripe. So a single sequential read would
result in same data being read twice.
This patch is to avoid chunk aligned read for degraded array. The
downside is to involve stripe cache which means associated CPU overhead
and extra memory copy.
Test Results:
Following test are done on a enterprise storage node with Seagate 6T SAS
drives and Xeon E5-2648L CPU (10 cores, 1.9Ghz), 10 disks MD RAID6 8+2,
chunk size 128 KiB.
I use FIO, using direct-io with various bs size, enough queue depth,
tested sequential and 100% random read against 3 array config:
1) optimal, as baseline;
2) degraded;
3) degraded with this patch.
Kernel version is 4.0-rc3.
Each individual test I only did once so there might be some variations,
but we just focus on big trend.
Sequential Read:
bs=(KiB) optimal(MiB/s) degraded(MiB/s) degraded-with-patch (MiB/s)
1024 1608 656 995
512 1624 710 956
256 1635 728 980
128 1636 771 983
64 1612 1119 1000
32 1580 1420 1004
16 1368 688 986
8 768 647 953
4 411 413 850
Random Read:
bs=(KiB) optimal(IOPS) degraded(IOPS) degraded-with-patch (IOPS)
1024 163 160 156
512 274 273 272
256 426 428 424
128 576 592 591
64 726 724 726
32 849 848 837
16 900 970 971
8 927 940 929
4 948 940 955
Some notes:
* In sequential + optimal, as bs size getting smaller, the FIO thread
become CPU bound.
* In sequential + degraded, there's big increase when bs is 64K and
32K, I don't have explanation.
* In sequential + degraded-with-patch, the MD thread mostly become CPU
bound.
If you want to we can discuss specific data point in those data. But in
general it seems with this patch, we have more predictable and in most
cases significant better sequential read performance when array is
degraded, and almost no noticeable impact on random read.
Performance is a complicated thing, the patch works well for this
particular configuration, but may not be universal. For example I
imagine testing on all SSD array may have very different result. But I
personally think in most cases IO bandwidth is more scarce resource than
CPU.
Signed-off-by: Eric Mei <eric.mei@seagate.com>
Signed-off-by: NeilBrown <neilb@suse.de>
|
|
The default setting of 256 stripe_heads is probably
much too small for many configurations. So it is best to make it
auto-configure.
Shrinking the cache under memory pressure is easy. The only
interesting part here is that we put a fairly high cost
('seeks') on shrinking the cache as the cost is greater than
just having to read more data, it reduces parallelism.
Growing the cache on demand needs to be done carefully. If we allow
fast growth, that can upset memory balance as lots of dirty memory can
quickly turn into lots of memory queued in the stripe_cache.
It is important for the raid5 block device to appear congested to
allow write-throttling to work.
So we only add stripes slowly. We set a flag when an allocation
fails because all stripes are in use, allocate at a convenient
time when that flag is set, and don't allow it to be set again
until at least one stripe_head has been released for re-use.
This means that a spurt of requests will only cause one stripe_head
to be allocated, but a steady stream of requests will slowly
increase the cache size - until memory pressure puts it back again.
It could take hours to reach a steady state.
The value written to, and displayed in, stripe_cache_size is
used as a minimum. The cache can grow above this and shrink back
down to it. The actual size is not directly visible, though it can
be deduced to some extent by watching stripe_cache_active.
Signed-off-by: NeilBrown <neilb@suse.de>
|
|
This allows us to easily add more (atomic) flags.
Signed-off-by: NeilBrown <neilb@suse.de>
|
|
Rather than adjusting max_nr_stripes whenever {grow,drop}_one_stripe()
succeeds, do it inside the functions.
Also choose the correct hash to handle next inside the functions.
This removes duplication and will help with future new uses of
{grow,drop}_one_stripe.
This also fixes a minor bug where the "md/raid:%md: allocate XXkB"
message always said "0kB".
Signed-off-by: NeilBrown <neilb@suse.de>
|
|
This is needed for future improvement to stripe cache management.
Signed-off-by: NeilBrown <neilb@suse.de>
|
|
Depending on the available coding we allow optimized rmw logic for write
operations. To support easier testing this patch allows manual control
of the rmw/rcw descision through the interface /sys/block/mdX/md/rmw_level.
The configuration can handle three levels of control.
rmw_level=0: Disable rmw for all RAID types. Hardware assisted P/Q
calculation has no implementation path yet to factor in/out chunks of
a syndrome. Enforcing this level can be benefical for slow CPUs with
hardware syndrome support and fast SSDs.
rmw_level=1: Estimate rmw IOs and rcw IOs. Execute rmw only if we will
save IOs. This equals the "old" unpatched behaviour and will be the
default.
rmw_level=2: Execute rmw even if calculated IOs for rmw and rcw are
equal. We might have higher CPU consumption because of calculating the
parity twice but it can be benefical otherwise. E.g. RAID4 with fast
dedicated parity disk/SSD. The option is implemented just to be
forward-looking and will ONLY work with this patch!
Signed-off-by: Markus Stockhausen <stockhausen@collogia.de>
Signed-off-by: NeilBrown <neilb@suse.de>
|
|
Glue it altogehter. The raid6 rmw path should work the same as the
already existing raid5 logic. So emulate the prexor handling/flags
and split functions as needed.
1) Enable xor_syndrome() in the async layer.
2) Split ops_run_prexor() into RAID4/5 and RAID6 logic. Xor the syndrome
at the start of a rmw run as we did it before for the single parity.
3) Take care of rmw run in ops_run_reconstruct6(). Again process only
the changed pages to get syndrome back into sync.
4) Enhance set_syndrome_sources() to fill NULL pages if we are in a rmw
run. The lower layers will calculate start & end pages from that and
call the xor_syndrome() correspondingly.
5) Adapt the several places where we ignored Q handling up to now.
Performance numbers for a single E5630 system with a mix of 10 7200k
desktop/server disks. 300 seconds random write with 8 threads onto a
3,2TB (10*400GB) RAID6 64K chunk without spare (group_thread_cnt=4)
bsize rmw_level=1 rmw_level=0 rmw_level=1 rmw_level=0
skip_copy=1 skip_copy=1 skip_copy=0 skip_copy=0
4K 115 KB/s 141 KB/s 165 KB/s 140 KB/s
8K 225 KB/s 275 KB/s 324 KB/s 274 KB/s
16K 434 KB/s 536 KB/s 640 KB/s 534 KB/s
32K 751 KB/s 1,051 KB/s 1,234 KB/s 1,045 KB/s
64K 1,339 KB/s 1,958 KB/s 2,282 KB/s 1,962 KB/s
128K 2,673 KB/s 3,862 KB/s 4,113 KB/s 3,898 KB/s
256K 7,685 KB/s 7,539 KB/s 7,557 KB/s 7,638 KB/s
512K 19,556 KB/s 19,558 KB/s 19,652 KB/s 19,688 Kb/s
Signed-off-by: Markus Stockhausen <stockhausen@collogia.de>
Signed-off-by: NeilBrown <neilb@suse.de>
|
|
expansion/resync can grab a stripe when the stripe is in batch list. Since all
stripes in batch list must be in the same state, we can't allow some stripes
run into expansion/resync. So we delay expansion/resync for stripe in batch
list.
Signed-off-by: Shaohua Li <shli@fusionio.com>
Signed-off-by: NeilBrown <neilb@suse.de>
|
|
If io error happens in any stripe of a batch list, the batch list will be
split, then normal process will run for the stripes in the list.
Signed-off-by: Shaohua Li <shli@fusionio.com>
Signed-off-by: NeilBrown <neilb@suse.de>
|
|
stripe cache is 4k size. Even adjacent full stripe writes are handled in 4k
unit. Idealy we should use big size for adjacent full stripe writes. Bigger
stripe cache size means less stripes runing in the state machine so can reduce
cpu overhead. And also bigger size can cause bigger IO size dispatched to under
layer disks.
With below patch, we will automatically batch adjacent full stripe write
together. Such stripes will be added to the batch list. Only the first stripe
of the list will be put to handle_list and so run handle_stripe(). Some steps
of handle_stripe() are extended to cover all stripes of the list, including
ops_run_io, ops_run_biodrain and so on. With this patch, we have less stripes
running in handle_stripe() and we send IO of whole stripe list together to
increase IO size.
Stripes added to a batch list have some limitations. A batch list can only
include full stripe write and can't cross chunk boundary to make sure stripes
have the same parity disks. Stripes in a batch list must be in the same state
(no written, toread and so on). If a stripe is in a batch list, all new
read/write to add_stripe_bio will be blocked to overlap conflict till the batch
list is handled. The limitations will make sure stripes in a batch list be in
exactly the same state in the life circly.
I did test running 160k randwrite in a RAID5 array with 32k chunk size and 6
PCIe SSD. This patch improves around 30% performance and IO size to under layer
disk is exactly 32k. I also run a 4k randwrite test in the same array to make
sure the performance isn't changed with the patch.
Signed-off-by: Shaohua Li <shli@fusionio.com>
Signed-off-by: NeilBrown <neilb@suse.de>
|
|
Track overwrite disk count, so we can know if a stripe is a full stripe write.
Signed-off-by: Shaohua Li <shli@fusionio.com>
Signed-off-by: NeilBrown <neilb@suse.de>
|
|
A freshly new stripe with write request can be batched. Any time the stripe is
handled or new read is queued, the flag will be cleared.
Signed-off-by: Shaohua Li <shli@fusionio.com>
Signed-off-by: NeilBrown <neilb@suse.de>
|
|
Use flex_array for scribble data. Next patch will batch several stripes
together, so scribble data should be able to cover several stripes, so this
patch also allocates scribble data for stripes across a chunk.
Signed-off-by: Shaohua Li <shli@fusionio.com>
Signed-off-by: NeilBrown <neilb@suse.de>
|
|
The patch makes 3 references to mddev->queue in the raid0 personality
conditional in order to allow for it to be accessed from dm-raid.
Mandatory, because md instances underneath dm-raid don't manage
a request queue of their own which'd lead to oopses without the patch.
Signed-off-by: Heinz Mauelshagen <heinzm@redhat.com>
Tested-by: Heinz Mauelshagen <heinzm@redhat.com>
Signed-off-by: NeilBrown <neilb@suse.de>
|
|
When md notices non-sync IO happening while it is trying
to resync (or reshape or recover) it slows down to the
set minimum.
The default minimum might have made sense many years ago
but the drives have become faster. Changing the default
to match the times isn't really a long term solution.
This patch changes the code so that instead of waiting until the speed
has dropped to the target, it just waits until pending requests
have completed.
This means that the delay inserted is a function of the speed
of the devices.
Testing shows that:
- for some loads, the resync speed is unchanged. For those loads
increasing the minimum doesn't change the speed either.
So this is a good result. To increase resync speed under such
loads we would probably need to increase the resync window
size.
- for other loads, resync speed does increase to a reasonable
fraction (e.g. 20%) of maximum possible, and throughput of
the load only drops a little bit (e.g. 10%)
- for other loads, throughput of the non-sync load drops quite a bit
more. These seem to be latency-sensitive loads.
So it isn't a perfect solution, but it is mostly an improvement.
Signed-off-by: NeilBrown <neilb@suse.de>
|
|
This option is not well justified and testing suggests that
it hardly ever makes any difference.
The comment suggests there might be a need to wait for non-resync
activity indicated by ->nr_waiting, however raise_barrier()
already waits for all of that.
So just remove it to simplify reasoning about speed limiting.
This allows us to remove a 'FIXME' comment from raid5.c as that
never used the flag.
Signed-off-by: NeilBrown <neilb@suse.de>
|
|
There is really no need for sync_min to be a multiple of
chunk_size, and values read from here often aren't.
That means you cannot read a value and expect to be able
to write it back later.
So remove the chunk_size check, and round down to a multiple
of 4K, to be sure everything works with 4K-sector devices.
Signed-off-by: NeilBrown <neilb@suse.de>
|
|
|
|
When "re-add" is writted to /sys/block/mdXX/md/dev-YYY/state,
the clustered md:
1. Sends RE_ADD message with the desc_nr. Nodes receiving the message
clear the Faulty bit in their respective rdev->flags.
2. The node initiating re-add, gathers the bitmaps of all nodes
and copies them into the local bitmap. It does not clear the bitmap
from which it is copying.
3. Initiating node schedules a md recovery to sync the devices.
Signed-off-by: Guoqing Jiang <gqjiang@suse.com>
Signed-off-by: Goldwyn Rodrigues <rgoldwyn@suse.com>
Signed-off-by: NeilBrown <neilb@suse.de>
|
|
This adds the capability of re-adding a failed disk by
writing "re-add" to /sys/block/mdXX/md/dev-YYY/state.
This facilitates adding disks which have encountered a temporary
error such as a network disconnection/hiccup in an iSCSI device,
or a SAN cable disconnection which has been restored. In such
a situation, you do not need to remove and re-add the device.
Writing re-add to the failed device's state would add it again
to the array and perform the recovery of only the blocks which
were written after the device failed.
This works for generic md, and is not related to clustering. However,
this patch is to ease re-add operations listed above in clustering
environments.
Signed-off-by: Goldwyn Rodrigues <rgoldwyn@suse.com>
Signed-off-by: NeilBrown <neilb@suse.de>
|
|
This adds "remove" capabilities for the clustered environment.
When a user initiates removal of a device from the array, a
REMOVE message with disk number in the array is sent to all
the nodes which kick the respective device in their own array.
This facilitates the removal of failed devices.
Signed-off-by: Goldwyn Rodrigues <rgoldwyn@suse.com>
Signed-off-by: NeilBrown <neilb@suse.de>
|
|
This is required by the clustering module (patches to follow) to
find the device to remove or re-add.
Signed-off-by: Goldwyn Rodrigues <rgoldwyn@suse.com>
Signed-off-by: NeilBrown <neilb@suse.de>
|
|
This export is required for clustering module in order to
co-ordinate remove/readd a rdev from all nodes.
Signed-off-by: Goldwyn Rodrigues <rgoldwyn@suse.com>
Signed-off-by: NeilBrown <neilb@suse.de>
|
|
Since the node num of md-cluster is from zero, and
cinfo->slot_number represents the slot num of dlm,
no need to check for equality.
Signed-off-by: Guoqing Jiang <gqjiang@suse.com>
Signed-off-by: Goldwyn Rodrigues <rgoldwyn@suse.com>
Signed-off-by: NeilBrown <neilb@suse.de>
|
|
Pull device mapper updates from Mike Snitzer:
- the most extensive changes this cycle are the DM core improvements to
add full blk-mq support to request-based DM.
- disabled by default but user can opt-in with CONFIG_DM_MQ_DEFAULT
- depends on some blk-mq changes from Jens' for-4.1/core branch so
that explains why this pull is built on linux-block.git
- update DM to use name_to_dev_t() rather than open-coding a less
capable device parser.
- includes a couple small improvements to name_to_dev_t() that offer
stricter constraints that DM's code provided.
- improvements to the dm-cache "mq" cache replacement policy.
- a DM crypt crypt_ctr() error path fix and an async crypto deadlock
fix
- a small efficiency improvement for DM crypt decryption by leveraging
immutable biovecs
- add error handling modes for corrupted blocks to DM verity
- a new "log-writes" DM target from Josef Bacik that is meant for file
system developers to test file system integrity at particular points
in the life of a file system
- a few DM log userspace cleanups and fixes
- a few Documentation fixes (for thin, cache, crypt and switch)
* tag 'dm-4.1-changes' of git://git.kernel.org/pub/scm/linux/kernel/git/device-mapper/linux-dm: (34 commits)
dm crypt: fix missing error code return from crypt_ctr error path
dm crypt: fix deadlock when async crypto algorithm returns -EBUSY
dm crypt: leverage immutable biovecs when decrypting on read
dm crypt: update URLs to new cryptsetup project page
dm: add log writes target
dm table: use bool function return values of true/false not 1/0
dm verity: add error handling modes for corrupted blocks
dm thin: remove stale 'trim' message documentation
dm delay: use msecs_to_jiffies for time conversion
dm log userspace base: fix compile warning
dm log userspace transfer: match wait_for_completion_timeout return type
dm table: fall back to getting device using name_to_dev_t()
init: stricter checking of major:minor root= values
init: export name_to_dev_t and mark name argument as const
dm: add 'use_blk_mq' module param and expose in per-device ro sysfs attr
dm: optimize dm_mq_queue_rq to _not_ use kthread if using pure blk-mq
dm: add full blk-mq support to request-based DM
dm: impose configurable deadline for dm_request_fn's merge heuristic
dm sysfs: introduce ability to add writable attributes
dm: don't start current request if it would've merged with the previous
...
|
|
Fix to return a negative error code from crypt_ctr()'s optional
parameter processing error path.
Signed-off-by: Wei Yongjun <yongjun_wei@trendmicro.com.cn>
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
|
|
I suspect this doesn't show up for most anyone because software
algorithms typically don't have a sense of being too busy. However,
when working with the Freescale CAAM driver it will return -EBUSY on
occasion under heavy -- which resulted in dm-crypt deadlock.
After checking the logic in some other drivers, the scheme for
crypt_convert() and it's callback, kcryptd_async_done(), were not
correctly laid out to properly handle -EBUSY or -EINPROGRESS.
Fix this by using the completion for both -EBUSY and -EINPROGRESS. Now
crypt_convert()'s use of completion is comparable to
af_alg_wait_for_completion(). Similarly, kcryptd_async_done() follows
the pattern used in af_alg_complete().
Before this fix dm-crypt would lockup within 1-2 minutes running with
the CAAM driver. Fix was regression tested against software algorithms
on PPC32 and x86_64, and things seem perfectly happy there as well.
Signed-off-by: Ben Collins <ben.c@servergy.com>
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
Cc: stable@vger.kernel.org
|
|
Commit 003b5c571 ("block: Convert drivers to immutable biovecs")
stopped short of changing dm-crypt to leverage the fact that the biovec
array of a bio will no longer be modified.
Switch to using bio_clone_fast() when cloning bios for decryption after
read.
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
|
|
Cryptsetup home page moved to GitLab.
Also remove link to abandonded Truecrypt page.
Signed-off-by: Milan Broz <gmazyland@gmail.com>
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
|
|
Introduce a new target that is meant for file system developers to test file
system integrity at particular points in the life of a file system. We capture
all write requests and associated data and log them to a separate device
for later replay. There is a userspace utility to do this replay. The
idea behind this is to give file system developers a tool to verify that
the file system is always consistent.
Signed-off-by: Josef Bacik <jbacik@fb.com>
Reviewed-by: Zach Brown <zab@zabbo.net>
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
|
|
Use the normal return values for bool functions.
Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
|
|
Add device specific modes to dm-verity to specify how corrupted
blocks should be handled. The following modes are defined:
- DM_VERITY_MODE_EIO is the default behavior, where reading a
corrupted block results in -EIO.
- DM_VERITY_MODE_LOGGING only logs corrupted blocks, but does
not block the read.
- DM_VERITY_MODE_RESTART calls kernel_restart when a corrupted
block is discovered.
In addition, each mode sends a uevent to notify userspace of
corruption and to allow further recovery actions.
The driver defaults to previous behavior (DM_VERITY_MODE_EIO)
and other modes can be enabled with an additional parameter to
the verity table.
Signed-off-by: Sami Tolvanen <samitolvanen@google.com>
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
|
|
Converting milliseconds to jiffies by "val * HZ / 1000" is technically
OK but msecs_to_jiffies(val) is the cleaner solution and handles all
corner cases correctly.
Signed-off-by: Nicholas Mc Guire <hofrat@osadl.org>
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
|
|
This fixes up a compile warning [-Wunused-but-set-variable] - given the
comment in userspace_set_region_sync() the non-reporting of errors is
intentional so the return value can be dropped to make gcc happy.
Also, fix typo in comment.
Signed-off-by: Nicholas Mc Guire <hofrat@osadl.org>
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
|
|
Return type of wait_for_completion_timeout() is unsigned long not int.
An appropriately named unsigned long is added and the assignment fixed.
Signed-off-by: Nicholas Mc Guire <hofrat@osadl.org>
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
|
|
If a device is used as the root filesystem, it can't be built
off of devices which are within the root filesystem (just like
command line arguments to root=). For this reason, Linux has a
pseudo-filesystem for root= and MD initialization (based on the
function name_to_dev_t) which handles different ways of specifying
devices including PARTUUID and major:minor.
Switch to using name_to_dev_t() in dm_get_device(). Rather than
having DM assume that all things which are not major:minor are paths in
an already-mounted filesystem, change dm_get_device() to first attempt
to look up the device in the filesystem, and if not found it will fall
back to using name_to_dev_t().
In terms of backwards compatibility, there are some cases where
behavior will be different:
- If you have a file in the current working directory named 1:2 and
you initialze DM there, then it will try to use that file rather
than the disk with that major:minor pair as a backing device.
- Similarly for other bdev types which name_to_dev_t() knows how to
interpret, the previous behavior was to repeatedly check for the
existence of the file (e.g., while waiting for rootfs to come up)
but the new behavior is to use the name_to_dev_t() interpretation.
For example, if you have a file named /dev/ubiblock0_0 which is
a symlink to /dev/sda3, but it is not yet present when DM starts
to initialize, then the name_to_dev_t() interpretation will take
precedence.
These incompatibilities would only show up in really strange setups
with bad practices so we shouldn't have to worry about them.
Signed-off-by: Dan Ehrenberg <dehrenberg@chromium.org>
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
|
|
Request-based DM's blk-mq support defaults to off; but a user can easily
change the default using the dm_mod.use_blk_mq module/boot option.
Also, you can check what mode a given request-based DM device is using
with: cat /sys/block/dm-X/dm/use_blk_mq
This change enabled further cleanup and reduced work (e.g. the
md->io_pool and md->rq_pool isn't created if using blk-mq).
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
|
|
dm_mq_queue_rq() is in atomic context so care must be taken to not
sleep -- as such GFP_ATOMIC is used for the md->bs bioset allocations
and dm-mpath's call to blk_get_request(). In the future the bioset
allocations will hopefully go away (by removing support for partial
completions of bios in a cloned request).
Also prepare for supporting DM blk-mq ontop of old-style request_fn
device(s) if a new dm-mod 'use_blk_mq' parameter is set. The kthread
will still be used to queue work if blk-mq is used ontop of old-style
request_fn device(s).
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
|
|
Commit e5863d9ad ("dm: allocate requests in target when stacking on
blk-mq devices") served as the first step toward fully utilizing blk-mq
in request-based DM -- it enabled stacking an old-style (request_fn)
request_queue ontop of the underlying blk-mq device(s). That first step
didn't improve performance of DM multipath ontop of fast blk-mq devices
(e.g. NVMe) because the top-level old-style request_queue was severely
limited by the queue_lock.
The second step offered here enables stacking a blk-mq request_queue
ontop of the underlying blk-mq device(s). This unlocks significant
performance gains on fast blk-mq devices, Keith Busch tested on his NVMe
testbed and offered this really positive news:
"Just providing a performance update. All my fio tests are getting
roughly equal performance whether accessed through the raw block
device or the multipath device mapper (~470k IOPS). I could only push
~20% of the raw iops through dm before this conversion, so this latest
tree is looking really solid from a performance standpoint."
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
Tested-by: Keith Busch <keith.busch@intel.com>
|
|
Otherwise, for sequential workloads, the dm_request_fn can allow
excessive request merging at the expense of increased service time.
Add a per-device sysfs attribute to allow the user to control how long a
request, that is a reasonable merge candidate, can be queued on the
request queue. The resolution of this request dispatch deadline is in
microseconds (ranging from 1 to 100000 usecs), to set a 20us deadline:
echo 20 > /sys/block/dm-7/dm/rq_based_seq_io_merge_deadline
The dm_request_fn's merge heuristic and associated extra accounting is
disabled by default (rq_based_seq_io_merge_deadline is 0).
This sysfs attribute is not applicable to bio-based DM devices so it
will only ever report 0 for them.
By allowing a request to remain on the queue it will block others
requests on the queue. But introducing a short dequeue delay has proven
very effective at enabling certain sequential IO workloads on really
fast, yet IOPS constrained, devices to build up slightly larger IOs --
yielding 90+% throughput improvements. Having precise control over the
time taken to wait for larger requests to build affords control beyond
that of waiting for certain IO sizes to accumulate (which would require
a deadline anyway). This knob will only ever make sense with sequential
IO workloads and the particular value used is storage configuration
specific.
Given the expected niche use-case for when this knob is useful it has
been deemed acceptable to expose this relatively crude method for
crafting optimal IO on specific storage -- especially given the solution
is simple yet effective. In the context of DM multipath, it is
advisable to tune this sysfs attribute to a value that offers the best
performance for the common case (e.g. if 4 paths are expected active,
tune for that; if paths fail then performance may be slightly reduced).
Alternatives were explored to have request-based DM autotune this value
(e.g. if/when paths fail) but they were quickly deemed too fragile and
complex to warrant further design and development time. If this problem
proves more common as faster storage emerges we'll have to look at
elevating a generic solution into the block core.
Tested-by: Shiva Krishna Merla <shivakrishna.merla@netapp.com>
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
|
|
Add DM_ATTR_RW() macro and establish .store method in dm_sysfs_ops.
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
|
|
Request-based DM's dm_request_fn() is so fast to pull requests off the
queue that steps need to be taken to promote merging by avoiding request
processing if it makes sense.
If the current request would've merged with previous request let the
current request stay on the queue longer.
Suggested-by: Jens Axboe <axboe@fb.com>
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
|
|
Commit 7eaceaccab ("block: remove per-queue plugging") didn't justify
DM's use of a 100ms delay; such an extended delay is a liability when
there is reason to re-kick the queue.
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
|
|
In request-based DM's dm_request_fn(), if blk_peek_request() returns
NULL just return. Avoids unnecessary blk_delay_queue().
Reported-by: Jens Axboe <axboe@fb.com>
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
|
|
On really fast storage it can be beneficial to delay running the
request_queue to allow the elevator more opportunity to merge requests.
Otherwise, it has been observed that requests are being sent to
q->request_fn much quicker than is ideal on IOPS-bound backends.
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
|
|
The old dm_request() method used for q->make_request_fn had a branch for
request-based DM support but it isn't needed given that
dm_init_request_based_queue() sets it to the standard blk_queue_bio()
anyway.
Cleanup dm_init_md_queue() to be DM device-type agnostic and have
dm_setup_md_queue() properly finish queue setup based on DM device-type
(bio-based vs request-based).
A followup block patch can be made to remove the export for
blk_queue_bio() now that DM no longer calls it directly.
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
|
|
Since commit 20d0189b1012a37d2533a87fb451f7852f2418d1
in v3.14-rc1 RAID0 has performed incorrect calculations
when the chunksize is not a power of 2.
This happens because "sector_div()" modifies its first argument, but
this wasn't taken into account in the patch.
So restore that first arg before re-using the variable.
Reported-by: Joe Landman <joe.landman@gmail.com>
Reported-by: Dave Chinner <david@fromorbit.com>
Fixes: 20d0189b1012a37d2533a87fb451f7852f2418d1
Cc: stable@vger.kernel.org (3.14 and later).
Signed-off-by: NeilBrown <neilb@suse.de>
|
|
Simon reported the md io stats accounting issue:
"
I'm seeing "iostat -x -k 1" print this after a RAID1 rebuild on 4.0-rc5.
It's not abnormal other than it's 3-disk, with one being SSD (sdc) and
the other two being write-mostly:
Device: rrqm/s wrqm/s r/s w/s rkB/s wkB/s avgrq-sz avgqu-sz await r_await w_await svctm %util
sda 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
sdb 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
sdc 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
md0 0.00 0.00 0.00 0.00 0.00 0.00 0.00 345.00 0.00 0.00 0.00 0.00 100.00
md2 0.00 0.00 0.00 0.00 0.00 0.00 0.00 58779.00 0.00 0.00 0.00 0.00 100.00
md1 0.00 0.00 0.00 0.00 0.00 0.00 0.00 12.00 0.00 0.00 0.00 0.00 100.00
"
The cause is commit "18c0b223cf9901727ef3b02da6711ac930b4e5d4" uses the
generic_start_io_acct to account the disk stats rather than the open code,
but it also introduced the increase to .in_flight[rw] which is needless to
md. So we re-use the open code here to fix it.
Reported-by: Simon Kirby <sim@hostway.ca>
Cc: <stable@vger.kernel.org> 3.19
Signed-off-by: Gu Zheng <guz.fnst@cn.fujitsu.com>
Signed-off-by: NeilBrown <neilb@suse.de>
|
|
DM multipath is the only caller of blk_lld_busy() -- which calls a
queue's lld_busy_fn hook. Request-based DM doesn't support stacking
multipath devices so there is no reason to register the lld_busy_fn hook
on a multipath device's queue using blk_queue_lld_busy().
As such, remove functions dm_lld_busy and dm_table_any_busy_target.
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
|
Linus Torvalds
Eric Mei
NeilBrown
Markus Stockhausen
shli@kernel.org
Heinz Mauelshagen
Goldwyn Rodrigues
Guoqing Jiang
Wei Yongjun
Ben Collins
Mike Snitzer
Milan Broz
Josef Bacik
Joe Perches
Sami Tolvanen
Nicholas Mc Guire
Dan Ehrenberg
Gu Zheng