A cryptic error code is printed when mounting a legacy dataset to a
non-existent mountpoint. This patch changes this behavior to print
"mount point '%s' does not exist", which is similar to the error
message printed when mounting procfs.
The single quotes were added to be consistent with the existing EBUSY
error message, which is the only difference between this error message
and the one that is printed when the same condition occurs when mounting
procfs.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#633
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>
When stdout is detected to be a tty use the number of columns
specified by the terminal. If that fails fall back to a default
80 column width. In the non-tty case allow for 999 column lines.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Caught by lint, this permission change was accidentally introduced
by commit 42cb3819f1. Restore the
correct permissions and while I'm at it add a missing whack-bang
to config/ltmain.sh.
lint: executable-not-elf-or-script: zpool_main.c zfs_main.c
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#620
Allow rigorous (and expensive) tx validation to be enabled/disabled
indepentantly from the standard zfs debugging. When enabled these
checks ensure that all txs are constructed properly and that a dbuf
is never dirtied without taking the correct tx hold.
This checking is particularly helpful when adding new dmu consumers
like Lustre. However, for established consumers such as the zpl
with no known outstanding tx construction problems this is just
overhead.
--enable-debug-dmu-tx - Enable/disable validation of each tx as
--disable-debug-dmu-tx it is constructed. By default validation
is disabled due to performance concerns.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
The following assertion is good to validate the correctness of
new DMU consumers, but it doesn't quite provide enough information.
Slightly rework the assertion so that when it is hit the actual
offending values will be included in the output.
SPLError: 4787:0:(dmu_tx.c:828:dmu_tx_dirty_buf())
ASSERTION(dn == NULL || dn->dn_assigned_txg == tx->tx_txg) failed
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Include the ZFS_META_RELEASE in the module load/unload messages
to more clearly indidcate exactly what version of ZFS has been
loaded.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Because the .zfs ctldir inodes are not backed by physical storage
they use a different create path which was not properly accounting
for them as used. This could result in ->nr_cached_objects()
returning 0 and cause a divide by zero error in prune_super().
In my option there's a kernel bug here too which allows this to
happen. They should either be checking for 0 or adding +1 like
they correctly do earlier in the function.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#617
Add support for the .zfs control directory. This was accomplished
by leveraging as much of the existing ZFS infrastructure as posible
and updating it for Linux as required. The bulk of the core
functionality is now all there with the following limitations.
*) The .zfs/snapshot directory automount support requires a 2.6.37
or newer kernel. The exception is RHEL6.2 which has backported
the d_automount patches.
*) Creating/destroying/renaming snapshots with mkdir/rmdir/mv
in the .zfs/snapshot directory works as expected. However,
this functionality is only available to root until zfs
delegations are finished.
* mkdir - create a snapshot
* rmdir - destroy a snapshot
* mv - rename a snapshot
The following issues are known defeciences, but we expect them to
be addressed by future commits.
*) Add automount support for kernels older the 2.6.37. This should
be possible using follow_link() which is what Linux did before.
*) Accessing the .zfs/snapshot directory via NFS is not yet possible.
The majority of the ground work for this is complete. However,
finishing this work will require resolving some lingering
integration issues with the Linux NFS kernel server.
*) The .zfs/shares directory exists but no futher smb functionality
has yet been implemented.
Contributions-by: Rohan Puri <rohan.puri15@gmail.com>
Contributiobs-by: Andrew Barnes <barnes333@gmail.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#173
Add a standard zio constructor and destructor. Normally, this is
done to reduce to cost of allocating a new structure by reducing
expensive operations such as memory allocations. However, in this
case none of the operations moved out of zio_create() were really
very expensive.
This change was principly made as a debug patch (and workaround)
for a zio_destroy() race. The is good evidence that zio_create()
is reinitializing a mutex which is really still in use by another
thread. This would completely explain the observed symptoms in
the issue report.
This patch doesn't fix the root cause of the race, but it should
make it less likely by only initializing the mutex once in the
constructor. Also, this particular flaw might have gone unnoticed
in other zfs implementations due to the specific implementation
details of Linux ticket spinlocks.
Once the real root cause is determined and resolved this change
can be safely reverted. Until then this should help workaround
the issue.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Issue #496
This patch was slightly flawed and allowed for zio->io_logical
to potentially not be reinitialized for a new zio. This could
lead to assertion failures in specific cases when debugging is
enabled (--enable-debug) and I/O errors are encountered. It
may also have caused problems when issues logical I/Os.
Since we want to make sure this workaround can be easily removed
in the future (when we have the real fix). I'm reverting this
change and applying a new version of the patch which includes
the zio->io_logical fix.
This reverts commit 2c6d0b1e07.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Issue #602
Issue #604
The xattr_resolve_name() helper function expects the registered
list of xattr handlers to be NULL terminated. This NULL was
accidentally missing which could result in a NULL dereference.
Interestingly this issue only manifested itself on certain 32-bit
systems. Presumably on 64-bit kernels we just always happen to
get lucky and the memory following the structure is zeroed.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Issue #594
The 'zfs list' and 'zpool list' commands output the message
'no datasets/pools available' to stdout. This should go to
stderr and only the available datasets/pools should go to
stdout. Returning nothing to stdout is expected behavior
when there is nothing to list.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#581
Add a SA interface which allows us to release the spill block
from a SA handle without destroying the handle. This is useful
because we can then ensure that a copy of the dirty spill block
is not made at sync time due to the extra hold. Susequent calls
to sa_update() or sa_lookup() with transparently refetch the
spill block dbuf from the ARC hash.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Add a standard zio constructor and destructor. Normally, this is
done to reduce to cost of allocating a new structure by reducing
expensive operations such as memory allocations. However, in this
case none of the operations moved out of zio_create() were really
very expensive.
This change was principly made as a debug patch (and workaround)
for a zio_destroy() race. The is good evidence that zio_create()
is reinitializing a mutex which is really still in use by another
thread. This would completely explain the observed symptoms in
the issue report.
This patch doesn't fix the root cause of the race, but it should
make it less likely by only initializing the mutex once in the
constructor. Also, this particular flaw might have gone unnoticed
in other zfs implementations due to the specific implementation
details of Linux ticket spinlocks.
Once the real root cause is determined and resolved this change
can be safely reverted. Until then this should help workaround
the issue.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Issue #496
Improve the distribution detection by moving the tests for
distribution specific files first. The Ubuntu and Debian
checks are left for last because they are the least likely
to be unique. This is particularly true in the case of Debian
since so many distributions are based on Debian.
Since this is currently only used to identify the correct
packaging method for this system the result in many instances
is simply cosmetic.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Some devices have exhibited sensitivity to the ending alignment of
partitions. In particular, even if the first partition begins at 1
MiB, we have seen many sd driver task abort errors with certain SSDs
if the first partition doesn't end on a 1 MiB boundary. This occurs
when the vdev label is read during pool creation or importation and
causes a delay of about 30 seconds per device. It can also be
simulated with dd when the pool isn't imported:
dd if=/dev/sda1 of=/dev/null bs=262144 count=1
For the record, this problem was observed with SMARTMOD
SG9XCA2E200GE01 200GB SSDs. Unfortunately I don't have a good
explanation for this behavior. It seems to have something to do with
highly fragmented single-sector requests being issued to the device,
which it may not support. With end-aligned partitions at least
page-sized requests were queued and issued to the driver according
to blktrace. In any case, aligning the partition end is a fairly
innocuous work-around, wasting at most 1 MiB of space.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#574
A private SA handle must be used to ensure we can drop the dbuf
hold on the spill block prior to calling dmu_tx_commit(). If we
call dmu_tx_commit() before sa_handle_destroy(), then our hold
will trigger a copy of the dbuf to be made. This is done to
prevent data from leaking in to the syncing txg. As a result
the original dirty spill block will remain cached.
Additionally, relying on the shared zp->z_sa_hdl is unsafe in
the xattr context because the znode may be asynchronously dropped
from the cache. It's far safer and simpler just to use a private
handle for xattrs. Plus any additional overhead is offset by
the avoidance of the previously mentioned memory copy.
These forever dirty buffers can be noticed in the arcstats under
the anon_size. On a quiescent system the value should be zero.
Without this fix and a SA xattr write workload you will see
anon_size increase. Eventually, if enough dirty data builds up
your system it will appear to hang. This occurs because the dmu
won't allow new txs to be assigned until that dirty data is
flushed, and it won't be because it's not part of an assigned tx.
As an aside, I typically see anon_size lurk around 16k so I think
there is another place in the code which needs a similar fix.
However, this value doesn't grow over time so it isn't critical.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Issue #503
Issue #513
Allow a source rpm to be rebuilt with debugging enabled. This
avoids the need to have to manually modify the spec file. By
default debugging is still largely disabled. To enable specific
debugging features use the following options with rpmbuild.
'--with debug' - Enables ASSERTs
# For example:
$ rpmbuild --rebuild --with debug zfs-modules-0.6.0-rc6.src.rpm
Additionally, ZFS_CONFIG has been added to zfs_config.h for
packages which build against these headers. This is critical
to ensure both zfs and the dependant package are using the same
prototype and structure definitions.
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>
When creating a new pool, make_root_vdev() calls check_in_use() to
ensure that none of the consituent disks are in use. If the disk
contains a valid vdev label it is read to retrieve the list of its
child vdevs and these are checked recursively. However, the
partitions stored in the vdev label my no longer exist, for example
if the partition table has since been altered. In any such case we
would want the pool creation to proceed, so this change removes the
check from check_slice() that returns an error if the device doesn't
exist. As an added assurance, the Solaris implementation also
returns sucess on ENOENT.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Export additional symbols to make use of the DMU's zero-copy
API. This allows external modules to move data in to and out of
the ARC without incurring the cost of a memory copy.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
New SSDs are now available which use an internal 8k block size.
To make sure ZFS can get the maximum performance out of these
devices we're increasing the maximum ashift to 13 (8KB).
This value is still small enough that we can fit 16 uberblocks
in the vdev ring label. However, I don't want to increase this
any futher or it will limit the ability the safely roll back a
pool to recover it.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#565
Resolves nfs-utils-1.0.x compatibility issue which requires
that the fsid be set in the export options.
exportfs: Warning: /tank/dir requires fsid= for NFS export
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#570
Exported the required symbols to make use of the DMU's zero-copy
API. This allows external modules to move data in to and out of
the ARC without incurring the cost of a memory copy.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
When configuring the spl debug log support use the provided wrapper
functions. This ensures that if --disable-debug-log was used when
buiding the spl the functions will have no effect.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
DISCARD (REQ_DISCARD, BLKDISCARD) is useful for thin provisioning.
It allows ZVOL clients to discard (unmap, trim) block ranges from
a ZVOL, thus optimizing disk space usage by allowing a ZVOL to
shrink instead of just grow.
We can't use zfs_space() or zfs_freesp() here, since these functions
only work on regular files, not volumes. Fortunately we can use the
low-level function dmu_free_long_range() which does exactly what we
want.
Currently the discard operation is not added to the log. That's not
a big deal since losing discard requests cannot result in data
corruption. It would however result in disk space usage higher than
it should be. Thus adding log support to zvol_discard() is probably
a good idea for a future improvement.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Currently only the (FALLOC_FL_PUNCH_HOLE) flag combination is
supported, since it's the only one that matches the behavior of
zfs_space(). This makes it pretty much useless in its current
form, but it's a start.
To support other flag combinations we would need to modify
zfs_space() to make it more flexible, or emulate the desired
functionality in zpl_fallocate().
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Issue #334
This isn't done on Solaris because on this OS zfs_space() can
only be called with an opened file handle. Since the addition of
zpl_truncate_range() this isn't the case anymore, so we need to
enforce access rights.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Issue #334
This operation allows "hole punching" in ZFS files. On Solaris this
is done via the vop_space() system call, which maps to the zfs_space()
function. So we just need to write zpl_truncate_range() as a wrapper
around zfs_space().
Note that this only works for regular files, not ZVOLs.
This is currently an insecure implementation without permission
checking, although this isn't that big of a deal since truncate_range()
isn't even callable from userspace.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Issue #334
The recent zvol improvements have changed default suggested alignment
for zvols from 512b (default) to 8k (zvol blocksize). Because of this
the zconfig.sh tests which create paritions are now generating a
warning about non-optimal alignments.
This change updates the need zconfig.sh tests such that a partition
will be properly aligned. In the process, it shifts from using the
sfdisk utility to the parted utility to create partitions. It also
moves the creation of labels, partitions, and filesystems in to
generic functions in common.sh.in.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Currently, the `zvol_threads` variable, which controls the number of worker
threads which process items from the ZVOL queues, is set to the number of
available CPUs.
This choice seems to be based on the assumption that ZVOL threads are
CPU-bound. This is not necessarily true, especially for synchronous writes.
Consider the situation described in the comments for `zil_commit()`, which is
called inside `zvol_write()` for synchronous writes:
> itxs are committed in batches. In a heavily stressed zil there will be a
> commit writer thread who is writing out a bunch of itxs to the log for a
> set of committing threads (cthreads) in the same batch as the writer.
> Those cthreads are all waiting on the same cv for that batch.
>
> There will also be a different and growing batch of threads that are
> waiting to commit (qthreads). When the committing batch completes a
> transition occurs such that the cthreads exit and the qthreads become
> cthreads. One of the new cthreads becomes he writer thread for the batch.
> Any new threads arriving become new qthreads.
We can easily deduce that, in the case of ZVOLs, there can be a maximum of
`zvol_threads` cthreads and qthreads. The default value for `zvol_threads` is
typically between 1 and 8, which is way too low in this case. This means
there will be a lot of small commits to the ZIL, which is very inefficient
compared to a few big commits, especially since we have to wait for the data
to be on stable storage. Increasing the number of threads will increase the
amount of data waiting to be commited and thus the size of the individual
commits.
On my system, in the context of VM disk image storage (lots of small
synchronous writes), increasing `zvol_threads` from 8 to 32 results in a 50%
increase in sequential synchronous write performance.
We should choose a more sensible default for `zvol_threads`. Unfortunately
the optimal value is difficult to determine automatically, since it depends
on the synchronous write latency of the underlying storage devices. In any
case, a hardcoded value of 32 would probably be better than the current
situation. Having a lot of ZVOL threads doesn't seem to have any real
downside anyway.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Fixes#392
The Linux block device queue subsystem exposes a number of configurable
settings described in Linux block/blk-settings.c. The defaults for these
settings are tuned for hard drives, and are not optimized for ZVOLs. Proper
configuration of these options would allow upper layers (I/O scheduler) to
take better decisions about write merging and ordering.
Detailed rationale:
- max_hw_sectors is set to unlimited (UINT_MAX). zvol_write() is able to
handle writes of any size, so there's no reason to impose a limit. Let the
upper layer decide.
- max_segments and max_segment_size are set to unlimited. zvol_write() will
copy the requests' contents into a dbuf anyway, so the number and size of
the segments are irrelevant. Let the upper layer decide.
- physical_block_size and io_opt are set to the ZVOL's block size. This
has the potential to somewhat alleviate issue #361 for ZVOLs, by warning
the upper layers that writes smaller than the volume's block size will be
slow.
- The NONROT flag is set to indicate this isn't a rotational device.
Although the backing zpool might be composed of rotational devices, the
resulting ZVOL often doesn't exhibit the same behavior due to the COW
mechanisms used by ZFS. Setting this flag will prevent upper layers from
making useless decisions (such as reordering writes) based on incorrect
assumptions about the behavior of the ZVOL.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
zvol_write() assumes that the write request must be written to stable storage
if rq_is_sync() is true. Unfortunately, this assumption is incorrect. Indeed,
"sync" does *not* mean what we think it means in the context of the Linux
block layer. This is well explained in linux/fs.h:
WRITE: A normal async write. Device will be plugged.
WRITE_SYNC: Synchronous write. Identical to WRITE, but passes down
the hint that someone will be waiting on this IO
shortly.
WRITE_FLUSH: Like WRITE_SYNC but with preceding cache flush.
WRITE_FUA: Like WRITE_SYNC but data is guaranteed to be on
non-volatile media on completion.
In other words, SYNC does not *mean* that the write must be on stable storage
on completion. It just means that someone is waiting on us to complete the
write request. Thus triggering a ZIL commit for each SYNC write request on a
ZVOL is unnecessary and harmful for performance. To make matters worse, ZVOL
users have no way to express that they actually want data to be written to
stable storage, which means the ZIL is broken for ZVOLs.
The request for stable storage is expressed by the FUA flag, so we must
commit the ZIL after the write if the FUA flag is set. In addition, we must
commit the ZIL before the write if the FLUSH flag is set.
Also, we must inform the block layer that we actually support FLUSH and FUA.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Currently the "sync=always" property works for regular ZFS datasets, but not
for ZVOLs. This patch remedies that.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Fixes#374.
Autoconf will fail to detect the ZoL libnvpair on systems that do not
implicitly link library runtime dependencies, which is anything that
has the GCC 4.5 DCO update.
Build libuutil before libnvpair, and put it on the the LDADD line of
the libnvpair automake template.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes: #560
The second argument of sops->show_options() was changed from a
'struct vfsmount *' to a 'struct dentry *'. Add an autoconf check
to detect the API change and then conditionally define the expected
interface. In either case we are only interested in the zfs_sb_t.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#549
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>
When using zfs to back a Lustre filesystem it's advantageous to
to store a fid with the object id in the directory zap. The only
technical impediment to doing this is that the zpl code expects
a single value in the zap per directory entry.
This change relaxes that requirement such that multiple entries
are allowed provided the first one is the object id. The zpl
code will just ignore additional entries. This allows the ZoL
count to mount datasets which are being used as Lustre server
backends.
Once the upstream feature flags support is merged in this change
should be updated to a read-only feature. Until this occurs
other zfs implementations will not be able to read the zfs
filesystems created by Lustre.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Export the zfs_attr_table symbol so it may be used by non-zpl
consumers which are still interested in writing a zpl compatible
dataset (e.g. Lustre).
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Since the zpios and potentially other ZFS tests use the
DMU_OST_OTHER type to label their datasets, the zpool and
zfs commands should gracefully handle this type when it is
encountered. This patch modifies the commands' behavior
to ignore any datasets with a dds_type of DMU_OST_OTHER.
Signed-off-by: Prakash Surya <surya1@llnl.gov>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#536
This change updates the rpm spec files to have strictly correct
package dependencies. That means a few things:
* The zfs-modules package is now tied to a specific build of
the spl-modules packages based on the kernel version. This
ensures that the correct spl-modules packages will always get
installed and not just the newest.
* The zfs package now requires both the zfs-modules and spl
packages. Thus a 'yum install zfs' will pull in the minimal
set of packages required for a functional system.
* The zfs-devel packages now require the zfs package to be
installed which is normal behavior for -devel packages.
* Remove the redundant distribution release extension. This
is already added once because it is part of the kernel package
release name.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
When the original build system code was added the release
component was accidentally omited from the development header
install path. This patch adds the missing path component so
it's always clear exactly what release your compiling against.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Commit zfsonlinux/zfs@57a4eddc4d
allows the bootfs property to be set on any pool, but does not
accommodate subsequent vdev changes. For example:
# zpool replace rpool /dev/sda /dev/sdb
operation not supported on this type of pool
property 'bootfs' is not supported on EFI labeled devices
For non-Solaris builds, disable the check that emits this error.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
These libraries, which are an artifact of the ZoL development
process, conflict with packages that are already in distribution:
* libspl: SPL Programming Language
* libavl: AVL for Linux
* libefi: GRUB
And these libraries are potential conflicts:
* libshare: the Linux Mount Manager
* libunicode: Perl and Python
Recompose these five ZoL components into the four libraries that are
conventionally provided by Solaris and FreeBSD systems:
+ libnvpair
+ libuutil
+ libzpool
+ libzfs
This change resolves the name conflict, makes ZoL more compatible
with existing software that uses autotools to detect ZFS, and allows
pkg-zfs to better reflect the official Debian kFreeBSD packaging.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes: #430
The vdev_is_bootable() restrictions are no longer necessary
with recent GRUB2 code. FreeBSD has implemented the same
change, except that I moved the Solaris comment to be inside
the #ifdef __sun__ block.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Issue #317
As described in Issue #458 and #258, unlinking large amounts of data
can cause the threads in the zio free wait queue to start spinning.
Reducing the number of z_fr_iss threads from a fixed value of 100 to 1
per cpu signficantly reduces contention on the taskq spinlock and
improves throughput.
Instrumenting the taskq code showed that __taskq_dispatch() can spend
a long time holding tq->tq_lock if there are a large number of threads
in the queue. It turns out the time spent in wake_up() scales
linearly with the number of threads in the queue. When a large number
of short work items are dispatched, as seems to be the case with
unlink, the worker threads drain the queue faster than the dispatcher
can fill it. They then all pile into the work wait queue to wait for
new work items. So if 100 threads are in the queue, wake_up() takes
about 100 times as long, and the woken threads have to spin until the
dispatcher releases the lock.
Reducing the number of threads helps with the symptoms, but doesn't
get to the root of the problem. It would seem that wake_up()
shouldn't scale linearly in time with queue depth, particularly if we
are only trying to wake up one thread. In that vein, I tried making
all of the waiting processes exclusive to prevent the scheduler from
iterating over the entire list, but I still saw the linear time
scaling. So further investigation is needed, but in the meantime
reducing the thread count is an easy workaround.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Issue #258
Issue #458
