7448 ZFS doesn't notice when disk vdevs have no write cache
illumos/illumos-gate@295438ba32295438ba32https://www.illumos.org/issues/7448
I built a SmartOS image with all the NVMe commits including 7372
(support NVMe volatile write cache) and repeated my dd testing:
> #!/bin/bash
> for i in `seq 1 1000`; do
> dd if=/dev/zero of=file00 bs=1M count=102400 oflag=sync &
> dd if=/dev/zero of=file01 bs=1M count=102400 oflag=sync &
> wait
> rm file00 file01
> done
>
Previously each dd command took ~145 seconds to finish, now it takes
~400 seconds.
Eventually I figured out it is 7372 that causes unnecessary
nvme_bd_sync() executions which wasted CPU cycles.
If a NVMe device doesn't support a write cache, the nvme_bd_sync function will
return ENOTSUP to indicate this to upper layers.
It seems this returned value is ignored by ZFS, and as such this bug is not
really specific to NVMe. In vdev_disk_io_start() ZFS sends the flush to the
disk driver (blkdev) with a callback to vdev_disk_ioctl_done(). As nvme filled
in the bd_sync_cache function pointer, blkdev will not return ENOTSUP, as the
nvme driver in general does support cache flush. Instead it will issue an
asynchronous flush to nvme and immediately return 0, and hence ZFS will not set
vdev_nowritecache here. The nvme driver will at some point process the cache
flush command, and if there is no write cache on the device it will return
ENOTSUP, which will be delivered to the vdev_disk_ioctl_done() callback. This
function will not check the error code and not set nowritecache.
The right place to check the error code from the cache flush is in
zio_vdev_io_assess(). This would catch both cases, synchronous and asynchronous
cache flushes. This would also be independent of the implementation detail that
some drivers can return ENOTSUP immediately.
Reviewed by: Dan Fields <dan.fields@nexenta.com>
Reviewed by: Alek Pinchuk <alek.pinchuk@nexenta.com>
Reviewed by: George Wilson <george.wilson@delphix.com>
Approved by: Dan McDonald <danmcd@omniti.com>
Author: Hans Rosenfeld <hans.rosenfeld@nexenta.com>
Obtained from: Illumos
7430 Backfill metadnode more intelligently
illumos/illumos-gate@af346df588af346df588https://www.illumos.org/issues/7430
Description and patch from brought over from the following ZoL commit: https://
github.com/zfsonlinux/zfs/commit/68cbd56e182ab949f58d004778d463aeb3f595c6
Only attempt to backfill lower metadnode object numbers if at least
4096 objects have been freed since the last rescan, and at most once
per transaction group. This avoids a pathology in dmu_object_alloc()
that caused O(N^2) behavior for create-heavy workloads and
substantially improves object creation rates. As summarized by
@mahrens in #4636:
"Normally, the object allocator simply checks to see if the next
object is available. The slow calls happened when dmu_object_alloc()
checks to see if it can backfill lower object numbers. This happens
every time we move on to a new L1 indirect block (i.e. every 32 *
128 = 4096 objects). When re-checking lower object numbers, we use
the on-disk fill count (blkptr_t:blk_fill) to quickly skip over
indirect blocks that don?t have enough free dnodes (defined as an L2
with at least 393,216 of 524,288 dnodes free). Therefore, we may
find that a block of dnodes has a low (or zero) fill count, and yet
we can?t allocate any of its dnodes, because they've been allocated
in memory but not yet written to disk. In this case we have to hold
each of the dnodes and then notice that it has been allocated in
memory.
The end result is that allocating N objects in the same TXG can
require CPU usage proportional to N^2."
Add a tunable dmu_rescan_dnode_threshold to define the number of
objects that must be freed before a rescan is performed. Don't bother
to export this as a module option because testing doesn't show a
compelling reason to change it. The vast majority of the performance
gain comes from limit the rescan to at most once per TXG.
Reviewed by: Alek Pinchuk <alek@nexenta.com>
Reviewed by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Approved by: Gordon Ross <gordon.w.ross@gmail.com>
Author: Ned Bass <bass6@llnl.gov>
Obtained from: Illumos
in place. To do per-cpu stats, convert all fields that previously were
maintained in the vmmeters that sit in pcpus to counter(9).
- Since some vmmeter stats may be touched at very early stages of boot,
before we have set up UMA and we can do counter_u64_alloc(), provide an
early counter mechanism:
o Leave one spare uint64_t in struct pcpu, named pc_early_dummy_counter.
o Point counter(9) fields of vmmeter to pcpu[0].pc_early_dummy_counter,
so that at early stages of boot, before counters are allocated we already
point to a counter that can be safely written to.
o For sparc64 that required a whole dummy pcpu[MAXCPU] array.
Further related changes:
- Don't include vmmeter.h into pcpu.h.
- vm.stats.vm.v_swappgsout and vm.stats.vm.v_swappgsin changed to 64-bit,
to match kernel representation.
- struct vmmeter hidden under _KERNEL, and only vmstat(1) is an exclusion.
This is based on benno@'s 4-year old patch:
https://lists.freebsd.org/pipermail/freebsd-arch/2013-July/014471.html
Reviewed by: kib, gallatin, marius, lidl
Differential Revision: https://reviews.freebsd.org/D10156
While the former name is easier to read, the "_flags" suffix has a special
meaning for loader(8) and, thus, it was impossible to set the knob via
loader.conf(5). The loader interpreted the setting as flags that should
be passed to a kernel module named "vfs.zfs.debug".
Discussed with: smh
MFC after: 2 weeks
When opening a vdev whose path is unknown, vdev_geom must find a geom
provider with a label whose guids match the desired vdev. However, due to
partitioning, it is possible that two non-synonomous providers will share
some labels. For example, if the first partition starts at the beginning of
the drive, then ada0 and ada0p1 will share the first label. More troubling,
if the last partition runs to the end of the drive, then ada0p3 and ada0
will share the last label. If vdev_geom opens ada0 when it should've opened
ada0p3, then the pool won't be readable. If it opens ada0 when it should've
opened ada0p1, then it will corrupt some other partition when it writes the
3rd and 4th labels.
The easiest way to reproduce this problem is to install a mirrored root pool
with the default partition layout, then swap the positions of the two boot
drives and reboot. Whether the bug manifests depends on the order in which
geom lists its providers, which is arbitrary.
Fix this situation by modifying the search algorithm to prefer geom
providers that have all four labels intact. If no such provider exists, then
open whichever provider has the most.
Reviewed by: mav
MFC after: 3 weeks
Sponsored by: Spectra Logic Corp
Differential Revision: https://reviews.freebsd.org/D10365
When a member of a RAIDZ has been replaced with a device smaller than the
original, then the top level vdev can report its expand size as 16.0E.
The reduced child asize causes the RAIDZ to have a vdev_asize lower than its
vdev_max_asize which then results in an underflow during the calculation of
the parents expand size.
Fix this by updating the vdev_asize if it shrinks, which is already
protected by a check against vdev_min_asize so should always be safe.
Also for RAIDZ vdevs, ensure that the sum of their child vdev_min_asize is
always greater than the parents vdev_min_size.
Fixes: https://www.illumos.org/issues/7885
MFC after: 2 weeks
Sponsored by: Multiplay
7603 xuio_stat_wbuf_* should be declared (void)
illumos/illumos-gate@99aa8b550599aa8b5505https://www.illumos.org/issues/7603
The funcs are declared k&r style, where the args are not specified:
void xuio_stat_wbuf_copied();
They should be declared to take no arguments:
void xuio_stat_wbuf_copied(void);
Need to change both .c and .h.
Author: Prashanth Sreenivasa <pks@delphix.com>
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Reviewed by: Paul Dagnelie <pcd@delphix.com>
Reviewed by: Robert Mustacchi <rm@joyent.com>
Approved by: Richard Lowe <richlowe@richlowe.net>
illumos/illumos-gate@8363e80ae7https://github.com/illumos/illumos-gate/commit/8363e80ae72609660f6090766ca8c2c18https://www.illumos.org/issues/7303
This change introduces a new weighting algorithm to improve metaslab selection.
The new weighting algorithm relies on the SPACEMAP_HISTOGRAM feature. As a result,
the metaslab weight now encodes the type of weighting algorithm used
(size-based vs segment-based).
This also introduce a new allocation tracing facility and two new dcmds to help
debug allocation problems. Each zio now contains a zio_alloc_list_t structure
that is populated as the zio goes through the allocations stage. Here's an
example of how to use the tracing facility:
> c5ec000::print zio_t io_alloc_list | ::walk list | ::metaslab_trace
MSID DVA ASIZE WEIGHT RESULT VDEV
- 0 400 0 NOT_ALLOCATABLE ztest.0a
- 0 400 0 NOT_ALLOCATABLE ztest.0a
- 0 400 0 ENOSPC ztest.0a
- 0 200 0 NOT_ALLOCATABLE ztest.0a
- 0 200 0 NOT_ALLOCATABLE ztest.0a
- 0 200 0 ENOSPC ztest.0a
1 0 400 1 x 8M 17b1a00 ztest.0a
> 1ff2400::print zio_t io_alloc_list | ::walk list | ::metaslab_trace
MSID DVA ASIZE WEIGHT RESULT VDEV
- 0 200 0 NOT_ALLOCATABLE mirror-2
- 0 200 0 NOT_ALLOCATABLE mirror-0
1 0 200 1 x 4M 112ae00 mirror-1
- 1 200 0 NOT_ALLOCATABLE mirror-2
- 1 200 0 NOT_ALLOCATABLE mirror-0
1 1 200 1 x 4M 112b000 mirror-1
- 2 200 0 NOT_ALLOCATABLE mirror-2
If the metaslab is using segment-based weighting then the WEIGHT column will
display the number of segments available in the bucket where the allocation
attempt was made.
Author: George Wilson <george.wilson@delphix.com>
Reviewed by: Alex Reece <alex@delphix.com>
Reviewed by: Chris Siden <christopher.siden@delphix.com>
Reviewed by: Dan Kimmel <dan.kimmel@delphix.com>
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Reviewed by: Paul Dagnelie <paul.dagnelie@delphix.com>
Reviewed by: Pavel Zakharov <pavel.zakharov@delphix.com>
Reviewed by: Prakash Surya <prakash.surya@delphix.com>
Reviewed by: Don Brady <don.brady@intel.com>
Approved by: Richard Lowe <richlowe@richlowe.net>
This way we can avoid blocking the whole queue in the low memory
situations. It's better to sacrifice some I/O performance by not doing
the aggregation than to add an indefinite wait for more memory.
Reviewed by: smh
MFC after: 2 weeks
Sponsored by: Panzura
Differential Revision: https://reviews.freebsd.org/D9999
As ZFS can request up to SPA_MAXBLOCKSIZE memory block e.g. during zfs recv,
update the threshold at which we start agressive reclamation to use
SPA_MAXBLOCKSIZE (16M) instead of the lower zfs_max_recordsize which
defaults to 1M.
PR: 194513
Reviewed by: avg, mav
MFC after: 1 month
Sponsored by: Multiplay
Differential Revision: https://reviews.freebsd.org/D10012
illumos/illumos-gate@6de76ce2a96de76ce2a9https://www.illumos.org/issues/7867
It seems that in the case where arc_hdr_free_pdata() sees HDR_L2_WRITING() we
would fail to update the ARC space statistics.
In the normal case those statistics are updated in arc_free_data_buf(). But in
the arc_hdr_free_on_write() path we don't do that.
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Reviewed by: Dan Kimmel <dan.kimmel@delphix.com>
Approved by: Dan McDonald <danmcd@omniti.com>
Author: Andriy Gapon <avg@FreeBSD.org>
MFC after: 10 days
illumos/illumos-gate@c5bde7273ec5bde7273ehttps://www.illumos.org/issues/7843
get_clones_stat() could be very slow if a snapshot has many (thousands) clones.
Clone names are added to an nvlist that's created with NV_UNIQUE_NAME.
So, each time a new name is appended to the list, the whole list is searched
linearly to see if that name is not already in the list. That results in the
quadratic complexity.
That should be easy to fix as we know in advance that we should not get any
duplicate names, so we can drop NV_UNIQUE_NAME when creating the list.
Reviewed by: Pavel Zakharov <pavel.zakharov@delphix.com>
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Approved by: Dan McDonald <danmcd@omniti.com>
Author: Andriy Gapon <avg@FreeBSD.org>
MFC after: 1 week
Sponsored by: ClusterHQ
For short transactions overhead of context switch can be too large.
Skipping it gives significant latency reduction. For large ones,
including multiple ZIOs, latency is less critical, while throughput
there may become limited by checksumming speed of single CPU core.
To get best of both cases, execute last ZIO directly from calling
thread context to save latency, while all others (if there are any)
enqueue to taskqueues in traditional way.
MFC after: 2 weeks
Sponsored by: iXsystems, Inc.
7570 tunable to allow zvol SCSI unmap to return on commit of txn to ZIL
illumos/illumos-gate@1c9272b8611c9272b861https://www.illumos.org/issues/7570
Based on the discovery that every unmap waits for the commit of the txn to the ZIL,
introducing a very high latency to unmap commands, this behavior was made into a
tunable zvol_unmap_sync_enabled and set to false. The net impact of this change is
that by default SCSI unmap commands will result in space being freed within the zvol
(today they are ignored and returned with good status). However, unlike the code
today, instead of 18+ms per unmap, they take about 30us.
With the testing done on NTFS against a Win2k12 target, the new behavior should work
seamlessly. Files on the zvol that have already been set with the zfree application
will continue to write 0's when deleted, and any new files created since zvol
creation will send unmap commands when deleted. This behavior exists today, but with
this change the unmap commands will be processed and result in reclaim of space.
Author: Stephen Blinick <stephen.blinick@delphix.com>
Reviewed by: Dan Kimmel <dan.kimmel@delphix.com>
Reviewed by: Matt Ahrens <mahrens@delphix.com>
Reviewed by: Steve Gonczi <steve.gonczi@delphix.com>
Reviewed by: Pavel Zakharov <pavel.zakharov@delphix.com>
Reviewed by: Saso Kiselkov <saso.kiselkov@nexenta.com>
Reviewed by: Yuri Pankov <yuri.pankov@nexenta.com>
Approved by: Robert Mustacchi <rm@joyent.com>
While there, make a change to not evict a first buffer outside the
requested eviciton range.
To do:
- give more consistent names to the size variables
- upstream to OpenZFS
PR: 216178
Reported by: lev
Tested by: lev
MFC after: 2 weeks
callout(9) prohibits callout functions from sleeping.
illumos mutexes are emulated using sx(9).
spa_deadman() calls vdev_deadman() and the latter acquires vq_lock.
As a result we can get a more confusing panic instead of a specific
panic or no panic:
sleepq_add: td 0xfffff80019669960 to sleep on wchan 0xfffff8001cff4d88 with sleeping prohibited
This change adds another level of indirection where the deadman
callout schedules spa_deadman() to be executed on taskqueue_thread.
While there, use callout_schedule(0 instead of callout_reset()
in spa_sync().
Discussed with: mav
MFC after: 1 week
Differential Revision: https://reviews.freebsd.org/D9762
6676 Race between unique_insert() and unique_remove() causes ZFS fsid change
illumos/illumos-gate@40510e8eba40510e8ebahttps://www.illumos.org/issues/6676
The fsid of zfs filesystems might change after reboot or remount. The problem seems to
be caused by a race between unique_insert() and unique_remove(). The unique_remove()
is called from dsl_dataset_evict() which is now an asynchronous thread. In a case the
dsl_dataset_evict() thread is very slow and calls unique_remove() too late we will end
up with changed fsid on zfs mount.
This problem is very likely caused by #5056.
Steps to Reproduce
Note: I'm able to reproduce this always on a single core (virtual) machine. On multicore
machines it is not so easy to reproduce.
# uname -a
SunOS openindiana 5.11 illumos-633aa80 i86pc i386 i86pc Solaris
# zfs create rpool/TEST
# FS=$(echo ::fsinfo | mdb -k | grep TEST | awk '{print $1}')
# echo $FS::print vfs_t vfs_fsid | mdb -k
vfs_fsid = {
vfs_fsid.val = [ 0x54d7028a, 0x70311508 ]
}
# zfs umount rpool/TEST
# zfs mount rpool/TEST
# FS=$(echo ::fsinfo | mdb -k | grep TEST | awk '{print $1}')
# echo $FS::print vfs_t vfs_fsid | mdb -k
vfs_fsid = {
vfs_fsid.val = [ 0xd9454e49, 0x6b36d08 ]
}
#
Impact
The persistent fsid (filesystem id) is essential for proper NFS functionality.
If the fsid of a filesystem changes on remount (or after reboot) the NFS
clients might not be able to automatically recover from such event and the
manual remount of the NFS filesystems on every NFS client might be needed.
Author: Josef 'Jeff' Sipek <josef.sipek@nexenta.com>
Reviewed by: Saso Kiselkov <saso.kiselkov@nexenta.com>
Reviewed by: Sanjay Nadkarni <sanjay.nadkarni@nexenta.com>
Reviewed by: Dan Vatca <dan.vatca@gmail.com>
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Reviewed by: George Wilson <george.wilson@delphix.com>
Reviewed by: Sebastien Roy <sebastien.roy@delphix.com>
Approved by: Robert Mustacchi <rm@joyent.com>
The difference of one was insignificant because zio_write_issue threads
ended up on the same run queues as other zio threads.
See sys/priority.h and sys/runq.h for more details.
Add a comment describing FreeBSD priority considerations and restore
the illumos variant of the code for comparison.
Obtained from: Panzura
MFC after: 2 weeks
Sponsored by: Panzura
The current code is written on top of GFS, a library with the generic
support for writing filesystems, which was ported from illumos.
Because of significant differences between illumos VFS and FreeBSD
VFS models, both the GFS and zfsctl code were heavily modified to
work on FreeBSD. Nonetheless, they still contain quite a few ugly
hacks and bugs.
This is a reimplementation of the zfsctl code where the VFS-specific
bits are written from scratch and only the code that interacts with
the rest of ZFS is reused.
Some highlights.
We use two types of nodes, static and on-demand. The static nodes
are used for permanent directories like .zfs, .zfs/snapshot, etc. The
on-demand nodes are used for ephemeral directories that act as snapshot
mount points.
Initially only static nodes are created. Their vnodes are instantiated
when they are looked up. The on-demand nodes and vnodes are instantiated
as needed and the nodes are destroyed as soon as the corresponding
vnodes are reclaimed.
We also try very hard to ensure that uncovered snapshot vnodes do not
linger. They are supposed to become inactive as soon as they are
uncovered and we try to recycle them immediately.
When a filesystem is unmounted all snapshots under .zfs are unmounted
first, then all vnodes are flushed and finally the static .zfs nodes
are destroyed.
There are some changes outside of zfsctl code too.
z_ctldir is never used directly (as it is an opaque pointer),
zfsctl_root() has to be used instead. The function returns a locked
vnode now, so it accepts a lock flags parameter. The function can
also fail now, e.g. during force unmounting, whereas previously it
was infallible.
zfsctl_root_lookup() is retired, instead of it VOP_LOOKUP() on the .zfs
vnode (obtained with zfsctl_root) is used.
Some ideas are picked from an independent work by will.
Reviewed by: asomers, smh
MFC after: 1 month
Relnotes: maybe
Differential Revision: https://reviews.freebsd.org/D7421
7504 kmem_reap hangs spa_sync and administrative tasks
illumos/illumos-gate@405a5a0f5chttps://github.com/illumos/illumos-gate/commit/405a5a0f5c3ab36cb76559467d1a62ba648bd80https://www.illumos.org/issues/7504
We see long spa_sync(). We are waiting to hold dp_config_rwlock for writer. Some
other thread holds dp_config_rwlock for reader, then calls arc_get_data_buf(),
which finds that arc_is_overflowing()==B_TRUE. So it waits (while holding
dp_config_rwlock for reader) for arc_reclaim_thread to signal arc_reclaim_waiters_cv.
Before signaling, arc_reclaim_thread does arc_kmem_reap_now(), which takes ~seconds.
Author: Matthew Ahrens <mahrens@delphix.com>
Reviewed by: George Wilson <george.wilson@delphix.com>
Reviewed by: Prakash Surya <prakash.surya@delphix.com>
Approved by: Dan McDonald <danmcd@omniti.com>
7500 Simplify dbuf_free_range by removing dn_unlisted_l0_blkid
illumos/illumos-gate@653af1b809653af1b809https://www.illumos.org/issues/7500
With the integration of:
commit 0f6d88aded0d165f5954688a9b13bac76c38da84
Author: Alex Reece <alex@delphix.com>
Date: Sat Jul 26 13:40:04 2014 -0800
4873 zvol unmap calls can take a very long time for larger datasets
the dnode's dn_bufs field was changed from a list to a tree. As a result,
the dn_unlisted_l0_blkid field is no longer necessary.
Author: Stephen Blinick <stephen.blinick@delphix.com>
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Reviewed by: Dan Kimmel <dan.kimmel@delphix.com>
Approved by: Gordon Ross <gordon.w.ross@gmail.com>
sys/cddl/contrib/opensolaris/uts/common/fs/zfs/zfs_vnops.c
* In zfs_freebsd_setattr, if the caller wants to set the birthtime,
set the bits that zfs_settattr expects
* In zfs_setattr, if XAT_CREATETIME is set, set xoa_createtime,
expected by zfs_xvattr_set. The two levels of indirection seem
excessive, but it minimizes diffs vs OpenZFS.
* In zfs_setattr, check for overflow of va_birthtime (from delphij)
* Remove red herring in zfs_getattr
sys/cddl/contrib/opensolaris/uts/common/sys/vnode.h
* Un-booby-trap some macros
New tests are under review at https://github.com/pjd/pjdfstest/pull/6
Reviewed by: avg
MFC after: 3 weeks
Sponsored by: Spectra Logic Corp
Differential Revision: https://reviews.freebsd.org/D9353
6569 large file delete can starve out write ops
illumos/illumos-gate@ff5177ee8bff5177ee8bhttps://www.illumos.org/issues/6569
The core issue I've found is that there is no throttle for how many
deletes get assigned to one TXG. As a results when deleting large files
we end up filling consecutive TXGs with deletes/frees, then write
throttling other (more important) ops.
There is an easy test case for this problem. Try deleting several
large files (at least 1/2 TB) while you do write ops on the same
pool. What we've seen is performance of these write ops (let's
call it sideload I/O) would drop to zero.
More specifically the problem is that dmu_free_long_range_impl()
can/will fill up all of the dirty data in the pool "instantly",
before many of the sideload ops can get in. So sideload
performance will be impacted until all the files are freed.
The solution we have tested at Nexenta (with positive results)
creates a relatively simple throttle for how many "free" ops we let
into one TXG.
However this solution exposes other problems that should also be
addressed. If we are to slow down freeing of data that means one
has to wait even longer (assuming vnode ref count of 1) to get shell
back after an rm or for NFS thread to finish the free-ing op.
To avoid this the proposed solution is to call zfs_inactive() async
for "large" files. Async freeing then begs for the reclaimed space
to be accounted for in the zpool's "freeing" prop.
The other issue with having a longer delete is the inability to
export/unmount for a longer period of time. The proposed solution
is to interrupt freeing of blocks when a fs is unmounted.
Author: Alek Pinchuk <alek@nexenta.com>
Reviewed by: Matt Ahrens <mahrens@delphix.com>
Reviewed by: Sanjay Nadkarni <sanjay.nadkarni@nexenta.com>
Reviewed by: Pavel Zakharov <pavel.zakharov@delphix.com>
Approved by: Dan McDonald <danmcd@omniti.com>
Reviewed by: avg
Differential Revision: D9008
At least on FreeBSD there are no legal way to access media or get its
size without opening device/provider first. Postponing this caching
allows to skip several disk seeks per ZVOL/snapshot during import.
For HDD pool with 1 ZVOL in dev mode with 1000 snapshots this reduces
pool import time from 40 to 10 seconds.
MFC after: 2 weeks
Sponsored by: iXsystems, Inc.
Original commit "7090 zfs should improve allocation order" declares alloc
queue sorted by time and offset. But in practice io_offset is always zero,
so sorting happened only by time, while order of writes with equal time was
completely random. On Illumos this did not affected much thanks to using
high resolution timestamps. On FreeBSD due to using much faster but low
resolution timestamps it caused bad data placement on disks, affecting
further read performance.
This change switches zio_timestamp_compare() from comparing uninitialized
io_offset to really populated io_bookmark values. I haven't decided yet
what to do with timestampts, but on simple tests this change gives the
same peformance results by just making code to work as declared.
MFC after: 1 week
Note: there was a merge conflict resolved by me.
illumos/illumos-gate@43297f973a43297f973ahttps://www.illumos.org/issues/3821
We recently had nodes with some of the latest zfs bits panic on us in a
rollback-heavy environment. The following is from my preliminary analysis:
Let's look at where we died:
> $C
ffffff01ea6b9a10 taskq_dispatch+0x3a(0, fffffffff7d20450, ffffff5551dea920, 1)
ffffff01ea6b9a60 zil_clean+0xce(ffffff4b7106c080, 7e0f1)
ffffff01ea6b9aa0 dsl_pool_sync_done+0x47(ffffff4313065680, 7e0f1)
ffffff01ea6b9b70 spa_sync+0x55f(ffffff4310c1d040, 7e0f1)
ffffff01ea6b9c20 txg_sync_thread+0x20f(ffffff4313065680)
ffffff01ea6b9c30 thread_start+8()
If we dig in we can find that this dataset corresponds to a zone:
> ffffff4b7106c080::print zilog_t zl_os->os_dsl_dataset->ds_dir->dd_myname
zl_os->os_dsl_dataset->ds_dir->dd_myname = [ "8ffce16a-13c2-4efa-a233-
9e378e89877b" ]
Okay so we have a null taskq pointer. That only happens during the calls to
zil_open and zil_close. If we poke around we can see that we're actually in
midst of a rollback:
> ::pgrep zfs | ::printf "0x%x %s\\n" proc_t . p_user.u_psargs
0xffffff43262800a0 zfs rollback zones/15714eb6-f5ea-469f-ac6d-
4b8ab06213c2@marlin_init
0xffffff54e22a1028 zfs rollback zones/8ffce16a-13c2-4efa-a233-
9e378e89877b@marlin_init
0xffffff4362f3a058 zfs rollback zones/0ddb8e49-ca7e-42e1-8fdc-
4ac4ba8fe9f8@marlin_init
0xffffff5748e8d020 zfs rollback zones/426357b5-832d-4430-953e-
10cd45ff8e9f@marlin_init
0xffffff436b867008 zfs rollback zones/8f36bf37-8a9c-4a44-995c-
6d1b2751e6f5@marlin_init
0xffffff4381ad4090 zfs rollback zones/6c8eca18-fbd6-46dd-ac24-
2ed45cd0da70@marlin_init
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Reviewed by: Dan Kimmel <dan.kimmel@delphix.com>
Reviewed by: Pavel Zakharov <pavel.zakharov@delphix.com>
Reviewed by: Andriy Gapon <avg@FreeBSD.org>
Approved by: Richard Lowe <richlowe@richlowe.net>
Author: George Wilson <george.wilson@delphix.com>
MFC after: 3 weeks
illumos/illumos-gate@90f2c094b390f2c094b3https://www.illumos.org/issues/7181
zfsvfs_setup() is called in both zfs_mount and zfs_resume_fs paths.
dmu_objset_set_user(zfsvfs->z_os, zfsvfs) is called early in zfsvfs_setup()
before the setup is actually completed,
thus an under-constructed zfsvfs becomes visible.
Additionally, there is nothing to serialize the two call paths. As a result two
threads can step on each other's toes.
assertion failed: zilog->zl_clean_taskq == NULL, file:
../../common/fs/zfs/zil.c, line: 1772
> $c
vpanic()
0xfffffffffbdf6928()
zil_open+0x45(ffffff1bbc5dd000, fffffffff7993880)
zfsvfs_setup+0x84(ffffffb378d77000, 0)
zfs_resume_fs+0x132(ffffffb378d77000, ffffffb37ddcf000)
zfs_ioc_rollback+0x96(ffffffb37ddcf000, ffffff01dcdc4cd0, ffffff01aa091000)
zfsdev_ioctl+0x215(10a00000000, 5a19, 80465f8, 100003, ffffff01ab318368,
ffffff0004b59e58)
cdev_ioctl+0x39(10a00000000, 5a19, 80465f8, 100003, ffffff01ab318368,
ffffff0004b59e58)
spec_ioctl+0x60(ffffff0197737700, 5a19, 80465f8, 100003,
ffffff01ab318368, ffffff0004b59e58)
fop_ioctl+0x55(ffffff0197737700, 5a19, 80465f8, 100003,
ffffff01ab318368, ffffff0004b59e58)
ioctl+0x9b(7, 5a19, 80465f8)
sys_syscall32+0x1f7()
> ffffff1bbc5dd000::print objset_t os_zil
os_zil = 0xffffff1c053cf7c0
> 0xffffff1c053cf7c0::print zilog_t zl_clean_taskq
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Approved by: Gordon Ross <gordon.w.ross@gmail.com>
Author: Andriy Gapon <andriy.gapon@clusterhq.com>
MFC after: 2 weeks
already free blocks
7199 dsl_dataset_rollback_sync may try to free already free blocks
7200 no blocks must be born in a txg after a snaphot is created
illumos/illumos-gate@bfaed0b91ebfaed0b91ehttps://www.illumos.org/issues/7199
dsl_dataset_rollback_sync may try to free already freed blocks when it calls
dsl_destroy_head_sync_impl to destroy a temporary clone.
That happens if a snapshot to which we are rolling back and from which the
clone is created has some ZIL records.
https://www.illumos.org/issues/7200
No new blocks must be born in a dataset in the same TXG after a snapshot of the
dataset is taken.
Those blocks would have the same blk_birth as the dataset's ds_prev_snap_txg
and as such they would be presumed to belong o the snapshot while in fact they
do not.
All the datasets must be clean before sync tasks are run, so the described
scenario may happen only if one of the sync tasks dirties the dataset and
another sync task takes its snapshot.
Then, there will be another sync pass because of the dirty data and the new
blocks will be born in the same TXG when the data is written out.
It seems that almost all of the existing sync tasks modify only MOS and do not
dirty any objsets.
The only exception that I've been able to identify so far is the rollback which
can modify an objset when it zeroes out the objset's ZIL.
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Reviewed by: Brad Lewis <brad.lewis@delphix.com>
Approved by: Gordon Ross <gordon.w.ross@gmail.com>
Author: Andriy Gapon <andriy.gapon@clusterhq.com>
MFC after: 3 weeks
and zfs_ioc_rename
illumos/illumos-gate@690041b9ca690041b9cahttps://www.illumos.org/issues/7180
If a filesystem is not unmounted while the rename is being performed, then, for
example, a concurrect zfs rollback may call zfs_suspend_fs followed by
zfs_resume_fs on the same filesystem.
The latter takes the filesystem's name as an argument. If the filesystem name
changes as a result of the rename, then dmu_objset_hold(osname, zfsvfs, &os)
call in zfs_resume_fs would fail resulting in a kernel panic.
So far I have been able to reproduce this problem on FreeBSD where zfs rename
has -u option that skips the unmounting before doing the renaming.
But I think that in theory the same problem can occur on illumos as well,
because the unmounting is done in userland before invoking the rename ioctl and
there could be a race with, e.g., zfs mount.
panic: solaris assert: dmu_objset_hold(osname, zfsvfs, &zfsvfs->z_os) == 0 (0x2
== 0x0), file: /usr/devel/svn/head/sys/cddl/contrib/opensolaris/uts/common/fs/
zfs/zfs_vfsops.c, line: 2210
KDB: stack backtrace:
db_trace_self_wrapper() at db_trace_self_wrapper+0x2b/frame 0xfffffe004df30710
vpanic() at vpanic+0x182/frame 0xfffffe004df30790
panic() at panic+0x43/frame 0xfffffe004df307f0
assfail3() at assfail3+0x2c/frame 0xfffffe004df30810
zfs_resume_fs() at zfs_resume_fs+0xb9/frame 0xfffffe004df30860
zfs_ioc_rollback() at zfs_ioc_rollback+0x61/frame 0xfffffe004df308a0
zfsdev_ioctl() at zfsdev_ioctl+0x65c/frame 0xfffffe004df30940
devfs_ioctl_f() at devfs_ioctl_f+0x156/frame 0xfffffe004df309a0
kern_ioctl() at kern_ioctl+0x246/frame 0xfffffe004df30a00
sys_ioctl() at sys_ioctl+0x171/frame 0xfffffe004df30ae0
amd64_syscall() at amd64_syscall+0x2db/frame 0xfffffe004df30bf0
Xfast_syscall() at Xfast_syscall+0xfb/frame 0xfffffe004df30bf0
Reviewed by: Matt Ahrens <mahrens@delphix.com>
Reviewed by: Pavel Zakharov <pavel.zakharov@delphix.com>
Approved by: Richard Lowe <richlowe@richlowe.net>
MFC after: 2 weeks
It was very wrong to look at the vnode and znode internals without
having locked the vnode first.
Reported by: pho
Tested by: pho
MFC after: 1 week
X-MFC with: r308887
The idea was to avoid a false assertion in zfs_lock, but it was
implemented very dangerously and incorrectly.
Reported by: pho
Tested by: pho
MFC after: 1 week
due to zl_itx_list_sz not updated when async itx'es upgraded to sync.
Actually because of other changes about that time zl_itx_list_sz is not
really required to implement the functionality, so this patch removes
some unneeded broken code and variables.
Original idea of zil_slog_limit was to reduce chance of SLOG abuse by
single heavy logger, that increased latency for other (more latency critical)
loggers, by pushing heavy log out into the main pool instead of SLOG. Beside
huge latency increase for heavy writers, this implementation caused double
write of all data, since the log records were explicitly prepared for SLOG.
Since we now have I/O scheduler, I've found it can be much more efficient
to reduce priority of heavy logger SLOG writes from ZIO_PRIORITY_SYNC_WRITE
to ZIO_PRIORITY_ASYNC_WRITE, while still leave them on SLOG.
Existing ZIL implementation had problem with space efficiency when it
has to write large chunks of data into log blocks of limited size. In some
cases efficiency stopped to almost as low as 50%. In case of ZIL stored on
spinning rust, that also reduced log write speed in half, since head had to
uselessly fly over allocated but not written areas. This change improves
the situation by offloading problematic operations from z*_log_write() to
zil_lwb_commit(), which knows real situation of log blocks allocation and
can split large requests into pieces much more efficiently. Also as side
effect it removes one of two data copy operations done by ZIL code WR_COPIED
case.
While there, untangle and unify code of z*_log_write() functions.
Also zfs_log_write() alike to zvol_log_write() can now handle writes crossing
block boundary, that may also improve efficiency if ZPL is made to do that.
Sponsored by: iXsystems, Inc.
not remove user-space visible fields from vm_cnt or all of the references to
cached pages from comments. Those changes will come later.)
Reviewed by: kib, markj
Tested by: pho
Sponsored by: Dell EMC Isilon
Differential Revision: https://reviews.freebsd.org/D8497
Before this an earlier writes to a ZVOL opened without FSYNC could get to
ZIL after later writes to the same ZVOL opened with FSYNC. Fix this by
replicating functionality of ZPL (zv_sync_cnt equivalent to z_sync_cnt),
marking all log records sync if anybody opened the ZVOL with FSYNC.
MFC after: 2 weeks
