Commit Graph

1861 Commits

Author SHA1 Message Date
Brian Behlendorf
218b4e0a76 Add zfs_sb_prune_aliases() function
For kernels which do not implement a per-suberblock shrinker,
those older than Linux 3.1, the shrink_dcache_parent() function
was used to attempt to reclaim dentries.  This was found not be
entirely reliable and could lead to performance issues on older
kernels running meta-data heavy workloads.

To address this issue a zfs_sb_prune_aliases() function has been
added to implement this functionality.  It relies on traversing
the list of znodes for a filesystem and adding them to a private
list with a reference held.  The private list can then be safely
walked outside the z_znodes_lock to prune dentires and drop the
last reference so the inode can be freed.

This provides the same synchronous behavior as the per-filesystem
shrinker and has the advantage of depending on only long standing
interfaces.

Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Tim Chase <tim@chase2k.com>
Closes #3501
2015-06-22 10:22:49 -07:00
Brian Behlendorf
4c6a700910 Increase the number of iput taskq threads
The number of threads in the iput taskq has been increased to speed
up the number of iputs which can be handled.  This has been observed
to improve the  meta data reclaim regardless of zfs_sb_prune()
implementation in use.

The taskq has also been renamed z_iput to for consistency with the
rest of the I/O pipeline taskqs which are all named z_*.

Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Tim Chase <tim@chase2k.com>
2015-06-22 10:22:10 -07:00
Matus Kral
57ae840077 Linux 4.1 compat: use read_iter() / write_iter()
Linux 3.15 commit torvalds/linux@293bc98 introduced two new methods.
The ->read_iter() and ->write_iter() methods were designed to replace
the ->aio_read() and ->aio_write() interfaces.  Both interfaces were
preserved for several kernel releases in order to migrate all existing
consumers to the new interfaces.  But as of Linux 4.1 the legacy
interface has been retired and the ZFS code must be updated to use
the new interfaces.

Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #3352
2015-06-18 12:06:59 -07:00
Sören Tempel
ec1c11d5ee Update dracut README
Include information about zfs-lib.sh.in and mention that it is possible
to set the bootfs attribute for an entire pool.

Signed-off-by: Sören Tempel <soeren+git@soeren-tempel.net>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #3109
2015-06-18 11:58:53 -07:00
Sören Tempel
01d01ef0ba Only source '/lib/dracut-lib.sh' if it wasn't so far
Signed-off-by: Sören Tempel <soeren+git@soeren-tempel.net>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Issue #3109
2015-06-17 16:46:37 -07:00
Sören Tempel
edbaf2455c Force export if it failed the first time
Signed-off-by: Sören Tempel <soeren+git@soeren-tempel.net>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Issue #3109
2015-06-17 16:46:04 -07:00
Sören Tempel
eda3d4e1de Refactor dracut module
Provide '/lib/dracut-zfs-lib.sh' with utility functions.

Signed-off-by: Sören Tempel <soeren+git@soeren-tempel.net>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Issue #3109
2015-06-17 16:45:18 -07:00
Turbo Fredriksson
036391c980 Additional SYSV init script fixes.
Use the 'mount' command instead of /proc/mounts to get a list of matching
filesystems.

This because /proc/mounts reports a pool with a space 'rpool 1' as
'rpool\0401'. The space is encoded as 3-digit octal which is legal.
However 'printf "%b"', which we use to filter out other illegal
characters (such as slash, space etc) can't properly interpret this
because it expects 4-digit octal. We get a  instead of the space
we expected. The correct value should have been 'rpool\00401' (note
the additional leading zero).

So use 'mount', which interprets all backslash-escapes correctly,
instead.

Signed-off-by: Turbo Fredriksson turbo@bayour.com
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #3488
2015-06-17 13:30:03 -07:00
Brian Behlendorf
5d6a460362 Unify mount and share for 'zfs create/clone'
Both the 'zfs create' and 'zfs clone' commands are expected to
automatically mount and share new filesystems.  Since this is common
functionality it has been moved in to a shared helper function.

Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #3459
2015-06-17 11:01:16 -07:00
Tim Chase
90947b2357 3.12 compat, NUMA-aware per-superblock shrinker
Kernels >= 3.12 have a NUMA-aware superblock shrinker which is used in
ZoL by zfs_sb_prune().  This patch calls the shrinker for each on-line
NUMA node in order that memory be freed for each one.

Signed-off-by: Tim Chase <tim@chase2k.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #3495
2015-06-17 10:43:13 -07:00
Hajo Möller
410921241d Add -y option to zpool iostat
sysstat's iostat omits the first report when the -y option is used.
This patch adds that functionality and omits the first report with
statistics since system boot.

Signed-off-by: Hajo Möller <dasjoe@gmail.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #3439
2015-06-17 10:39:20 -07:00
Brian Behlendorf
8e70975f90 Wait interruptibly in prefetch thread
The Linux kernel watchdog will automatically dump a backtrace for
any process while sleeps for over 120s in an uninterruptible state.

The solution is for the prefetch thread to sleep in an interruptible
state.  The way the existing code was written this is safe because
when woken it will always reevaluate its conditional.  As a general
rule it is preferable to sleep in an interruptible when possible.

Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #3450
Closes #3402
2015-06-16 16:18:11 -07:00
Brian Behlendorf
b64ccd6c52 Rename cv_wait_interruptible() to cv_wait_sig()
This is the counterpart to zfsonlinux/spl@2345368 which replaces the
cv_wait_interruptible() function with cv_wait_sig().  There is no
functional change to patch merely brings the function names in to
sync to maximize portability.

Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Issue #3450
Issue #3402
2015-06-11 10:50:47 -07:00
Brian Behlendorf
06358ea16e Merge branch 'lock-contention-on-arcs_mtx-final'
Signed-off-by: Tim Chase <tim@chase2k.com>
Signed-off-by: Brian Behlendorf
Closes #3115
Closes #3481
2015-06-11 10:27:48 -07:00
Tim Chase
121b3cae74 Increase arc_c_min to allow safe operation of arc_adapt()
ZoL had lowered the minimum ARC size to 4MiB to better accommodate tiny
systems such as the raspberry pi, however, as of addition of large block
support, the arc_adapt() function depends on arc_c being >= 32MiB (2 *
SPA_MAXBLOCKSIZE).

This patch raises the minimum ARC size to 32MiB and adds a VERIFY test
to arc_adapt() for future-proofing.

Signed-off-by: Tim Chase <tim@chase2k.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
2015-06-11 10:27:25 -07:00
Brian Behlendorf
f604673836 Make arc_prune() asynchronous
As described in the comment above arc_adapt_thread() it is critical
that the arc_adapt_thread() function never sleep while holding a hash
lock.  This behavior was possible in the Linux implementation because
the arc_prune() logic was implemented to be synchronous.  Under
illumos the analogous dnlc_reduce_cache() function is asynchronous.

To address this the arc_do_user_prune() function is has been reworked
in to two new functions as follows:

* arc_prune_async() is an asynchronous implementation which dispatches
the prune callback to be run by the system taskq.  This makes it
suitable to use in the context of the arc_adapt_thread().

* arc_prune() is a synchronous implementation which depends on the
arc_prune_async() implementation but blocks until the outstanding
callbacks complete.  This is used in arc_kmem_reap_now() where it
is safe, and expected, that memory will be freed.

This patch additionally adds the zfs_arc_meta_strategy module option
while allows the meta reclaim strategy to be configured.  It defaults
to a balanced strategy which has been proved to work well under Linux
but the illumos meta-only strategy can be enabled.

Signed-off-by: Tim Chase <tim@chase2k.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
2015-06-11 10:27:25 -07:00
Brian Behlendorf
c5528b9ba6 Use taskq_wait_outstanding() function
Replace taskq_wait() with taskq_wait_oustanding().  This way callers
will only block until previously submitted tasks have been completed.
This was the previous behavior of task_wait() prior to the introduction
of taskq_wait_outstanding() so this isn't really a functionalty change
for these callers.

Signed-off-by: Tim Chase <tim@chase2k.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
2015-06-11 10:27:25 -07:00
Brian Behlendorf
4f34bd9792 Add taskq_wait_outstanding() function
SPL commit behlendorf/spl@9cef1b5 adds the taskq_wait_outstanding()
interface.  See the commit log for the full justification for this
addition.  This patch adds the required user space counterpart.

Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Tim Chase <tim@chase2k.com>
2015-06-11 10:27:25 -07:00
Prakash Surya
ca0bf58d65 Illumos 5497 - lock contention on arcs_mtx
Reviewed by: George Wilson <george.wilson@delphix.com>
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Reviewed by: Richard Elling <richard.elling@richardelling.com>
Approved by: Dan McDonald <danmcd@omniti.com>

Porting notes and other significant code changes:

The illumos 5368 patch (ARC should cache more metadata), which
was never picked up by ZoL, is mostly reverted by this patch.

Since ZoL relies on the kernel asynchronously calling the shrinker to
actually reap memory, the shrinker wakes up arc_reclaim_waiters_cv every
time it runs.

The arc_adapt_thread() function no longer calls arc_do_user_evicts()
since the newly-added arc_user_evicts_thread() calls it periodically.

Notable conflicting ZoL commits which conflicted with this patch or
whose effects are either duplicated or un-done by this patch:

    302f753 - Integrate ARC more tightly with Linux
    39e055c - Adjust arc_p based on "bytes" in arc_shrink
    f521ce1 - Allow "arc_p" to drop to zero or grow to "arc_c"
    77765b5 - Remove "arc_meta_used" from arc_adjust calculation
    94520ca - Prune metadata from ghost lists in arc_adjust_meta

Trace support for multilist_insert() and multilist_remove() has been
added and produces the following output:

    fio-12498 [077] .... 112936.448324: zfs_multilist__insert: ml { offset 240 numsublists 80 sublistidx 63 }
    fio-12498 [077] .... 112936.448347: zfs_multilist__remove: ml { offset 240 numsublists 80 sublistidx 29 }

The following arcstats have been removed:

    recycle_miss - Used by arcstat.py and arc_summary.py, both of which
    have been updated appropriately.

    l2_writes_hdr_miss

The following arcstats have been added:

    evict_not_enough - Number of times arc_evict_state() was unable to
    evict enough buffers to reach its target amount.

    evict_l2_skip - Number of times arc_evict_hdr() skipped eviction
    because it was being written to the l2arc.

    l2_writes_lock_retry - Replaces l2_writes_hdr_miss.  Number of times
    l2arc_write_done() failed to acquire hash_lock (and re-tries).

    arc_meta_min - Shows the value of the zfs_arc_meta_min module
    parameter (see below).

The "index" column of the "dbuf" kstat has been removed since it doesn't
have a direct analog in the new multilist scheme.  Additional multilist-
related stats could be added in the future but would likely require
extensions to the mulilist API.

The following module parameters have been added:

    zfs_arc_evict_batch_limit - Number of ARC headers to free per sub-list
    before moving on to the next sub-list.

    zfs_arc_meta_min - Enforce a floor on the amount of metadata in
    the ARC.

    zfs_arc_num_sublists_per_state - Number of multilist sub-lists per
    ARC state.

    zfs_arc_overflow_shift - Controls amount by which the ARC must exceed
    the target size to be considered "overflowing".

Ported-by: Tim Chase <tim@chase2k.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov
2015-06-11 10:27:25 -07:00
Chris Williamson
b9541d6b7d Illumos 5408 - managing ZFS cache devices requires lots of RAM
5408 managing ZFS cache devices requires lots of RAM
Reviewed by: Christopher Siden <christopher.siden@delphix.com>
Reviewed by: George Wilson <george.wilson@delphix.com>
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Reviewed by: Don Brady <dev.fs.zfs@gmail.com>
Reviewed by: Josef 'Jeff' Sipek <josef.sipek@nexenta.com>
Approved by: Garrett D'Amore <garrett@damore.org>

Porting notes:

Due to the restructuring of the ARC-related structures, this
patch conflicts with at least the following existing ZoL commits:

    6e1d7276c9
    Fix inaccurate arcstat_l2_hdr_size calculations

        The ARC_SPACE_HDRS constant no longer exists and has been
        somewhat equivalently replaced by HDR_L2ONLY_SIZE.

    e0b0ca983d
    Add visibility in to cached dbufs

        The new layering of l{1,2}arc_buf_hdr_t within the arc_buf_hdr
        struct requires additional structure member names to be used
        when referencing the inner items.  Also, the presence of L1 or L2
        inner member is indicated by flags using the new HDR_HAS_L{1,2}HDR
        macros.

Ported by: Tim Chase <tim@chase2k.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
2015-06-11 10:27:25 -07:00
George Wilson
2a4324141f Illumos 5369 - arc flags should be an enum
5369 arc flags should be an enum
5370 consistent arc_buf_hdr_t naming scheme
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Reviewed by: Alex Reece <alex.reece@delphix.com>
Reviewed by: Sebastien Roy <sebastien.roy@delphix.com>
Reviewed by: Richard Elling <richard.elling@richardelling.com>
Approved by: Richard Lowe <richlowe@richlowe.net>

Porting notes:

ZoL has moved some ARC definitions into arc_impl.h.

Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Ported by: Tim Chase <tim@chase2k.com>
2015-06-11 10:27:25 -07:00
Tim Chase
ad4af89561 Partially revert "Add ddt, ddt_entry, and l2arc_hdr caches"
This reverts only the l2arc_hdr part of commit
ecf3d9b8e6 in preparation for the illumos
5497 "lock contention on arcs_mtx" patch which does the same thing
but uses the newer two-level ARC structure following the Illumos 5408
"managing ZFS cache devices requires lots of RAM" patch.

Signed-off-by: Tim Chase <tim@chase2k.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
2015-06-11 10:27:25 -07:00
Tim Chase
97639d0a52 Revert "Allow arc_evict_ghost() to only evict meta data"
Illumos 5497 "lock contention on arcs_mtx" reworks eviction and obviates
the need for this.

Signed-off-by: Tim Chase <tim@chase2k.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
2015-06-11 10:27:25 -07:00
Tim Chase
f6b3b1f5d6 Revert "fix l2arc compression buffers leak"
This reverts commit 037763e44e in
preparation for the illumos 5497 "lock contention on arcs_mtx" patch
which includes a fix for this very problem.

ZoL had picked up a subset of the illumos 5497 patch to deal with the
l2arc compression buffer leak.

Signed-off-by: Tim Chase <tim@chase2k.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
2015-06-11 10:27:24 -07:00
Tim Chase
7807028ccd Revert "arc_evict, arc_evict_ghost: reduce stack usage using kmem_zalloc"
This reverts commit 16fcdea363 in preparation
for the illumos 5497 "lock contention on arcs_mtx" patch which eliminates
"marker" within the ARC code.

Signed-off-by: Tim Chase <tim@chase2k.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
2015-06-11 10:27:24 -07:00
Brian Behlendorf
44de2f02d6 Remove unused variable in vdev_add_child()
Commit c3520e7 restructured vdev_add_child() in such a way that
the spa variable was unused during non-debug builds.  This is
consistent with the upstream illumos code but because ZoL, unlike
illumos, is built with all compiler warnings enabled this causes
a legitimate warning.  Revert this hunk of the patch to keep the
build clean.

Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Issue #3432
2015-06-11 10:22:38 -07:00
Matthew Ahrens
c3520e7f1f Illumos 5818 - zfs {ref}compressratio is incorrect with 4k sector size
5818 zfs {ref}compressratio is incorrect with 4k sector size
Reviewed by: Alex Reece <alex@delphix.com>
Reviewed by: George Wilson <george@delphix.com>
Reviewed by: Richard Elling <richard.elling@richardelling.com>
Reviewed by: Steven Hartland <killing@multiplay.co.uk>
Approved by: Albert Lee <trisk@omniti.com>

References:
  https://www.illumos.org/issues/5818
  https://github.com/illumos/illumos-gate/commit/81cd5c5

Ported-by: Don Brady <don.brady@intel.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #3432
2015-06-10 16:24:01 -07:00
Arne Jansen
9c43027b3f Illumos 5269 - zpool import slow
5269 zpool import slow
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Reviewed by: George Wilson <george@delphix.com>
Reviewed by: Dan McDonald <danmcd@omniti.com>
Approved by: Dan McDonald <danmcd@omniti.com>

References:
  https://www.illumos.org/issues/5269
  https://github.com/illumos/illumos-gate/commit/12380e1e

Ported-by: DHE <git@dehacked.net>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #3396
2015-06-09 13:48:02 -07:00
Turbo Fredriksson
d050c627b5 Improve on the ZFS events documentation
* Add information about the 'zpool events' command in zpool(8).
* More events and payloads defined in zfs-events(5).
* I/O Stages and I/O Flags sections added.
* Remove unused legacy "zio_deadline" payload define.

Signed-off-by: Turbo Fredriksson <turbo@bayour.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #3467
2015-06-09 11:19:19 -07:00
Ned Bass
5f8e1e8505 dmu_objset_userquota_get_ids uses dn_bonus unsafely
The function dmu_objset_userquota_get_ids() checks and uses dn->dn_bonus
outside of dn_struct_rwlock. If the dnode is being freed then the bonus
dbuf may be in the process of getting evicted. In this case there is a
race that may cause dmu_objset_userquota_get_ids() to access the dbuf
after it has been destroyed. To prevent this, ensure that when we are
using the bonus dbuf we are either holding a reference on it or have
taken dn_struct_rwlock.

Signed-off-by: Ned Bass <bass6@llnl.gov>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #3443
2015-06-05 12:41:17 -07:00
Ned Bass
d617648c7f dbuf_try_add_ref minor bug fixes
- Don't check db->bb_blkid, but use the blkid argument instead.
  Checking db->db_blkid may be unsafe since we doesn't yet have a
  hold on the dbuf so its validity is unknown.

- Call mutex_exit() on found_db, not db, since it's not certain that
  they point to the same dbuf, and the mutex was taken on found_db.

Signed-off-by: Ned Bass <bass6@llnl.gov>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Issue #3443
2015-06-05 12:40:38 -07:00
Turbo Fredriksson
4f38c25910 SYSV init script fixes.
* Change the order of the function library check/load.
  Redhat based system _can_ have a /lib/lsb/init-functions file (from
  the redhat-lsb-core package), but it's only partially what we can use.
  Instead, look for that file last, giving the script a chance to catch
  the 'real' distribution file.
* Filter out dashes and dots in dataset name in read_mtab().
* Get rid of 'awk' entirely. This is usually in /usr, which might not
  be availible.
* Get rid of the 'find /dev/disk/by-*' (find is on /usr, which might not
  be availible). Instead use echo in a for loop.
* Rebuild scripts if any of the *.in files changed.
* Move the sed part that filters out duplicates inside the check fo
  valid variable.

Signed-off-by: Turbo Fredriksson turbo@bayour.com
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #3463
Closes #3457
2015-06-05 12:35:39 -07:00
Turbo Fredriksson
2a34db1bdb Base init scripts for SYSV systems
* Based on the init scripts included with Debian GNU/Linux, then take code
  from the already existing ones, trying to merge them into one set of
  scripts that will work for 'everyone' for better maintainability.
  * Add configurable variables to control the workings of the init scripts:
    * ZFS_INITRD_PRE_MOUNTROOT_SLEEP
      Set a sleep time before we load the module (used primarily by initrd
      scripts to allow for slower media (such as  USB devices etc) to be
      availible before we load the zfs module).
    * ZFS_INITRD_POST_MODPROBE_SLEEP
      Set a timed sleep in the initrd to after the load of the zfs module.
    * ZFS_INITRD_ADDITIONAL_DATASETS
      To allow for mounting additional datasets in the initrd. Primarily used
      in initrd scripts to allow for when filesystem needed to boot (such as
      /usr, /opt, /var etc) isn't directly under the root dataset.
    * ZFS_POOL_EXCEPTIONS
      Exclude pools from being imported (in the initrd and/or init scripts).
    * ZFS_DKMS_ENABLE_DEBUG, ZFS_DKMS_ENABLE_DEBUG_DMU_TX, ZFS_DKMS_DISABLE_STRIP
      Set to control how dkms should build the dkms packages.
    * ZPOOL_IMPORT_PATH
      Set path(s) where "zpool import" should import pools from.
      This was previously the job of "USE_DISK_BY_ID" (which is still used
      for backwards compatibility) but was renamed to allow for better
      control of import path(s).
      * If old USE_DISK_BY_ID is set, but not new ZPOOL_IMPORT_PATH, then we
        set ZPOOL_IMPORT_PATH to sane defaults just to be on the safe side.
    * ZED_ARGS
      To allow for local options to zed without having to change the init script.
  * The import function, do_import(), imports pools by name instead of '-a'
    for better control of pools to import and from where.
    * If USE_DISK_BY_ID is set (for backwards compatibility), but isn't 'yes'
      then ignore it.
    * If pool(s) isn't found with a simple "zpool import" (seen it happen),
      try looking for them in /dev/disk/by-id (if it exists). Any duplicates
      (pools found with both commands) is filtered out.
      * IF we have found extra pool(s) this way, we must force USE_DISK_BY_ID
        so that the first, simple "zpool import $pool" is able to find it.
    * Fallback on importing the pool using the cache file (if it exists) only
      if 'simple' import (either with ZPOOL_IMPORT_PATH or the 'built in'
      defaults) didn't work.
  * The export function, do_export(), will export all pools imported, EXCEPT
    the root pool (if there is one).
  * ZED script from the Debian GNU/Linux packages added.
    * Refreshed ZED init script from behlendorf@5e7a660 to be portable so it
      may be used on both LSB and Redhat style systems.
    * If there is no pool(s) imported and zed successfully shut down, we will
      unload the zfs modules.
  * The function library file for the ZoL init script is installed as
    /etc/init.d/zfs-functions.
  * The four init scripts, the /etc/{defaults,sysconfig,conf.d}/zfs config file
    as well as the common function library is tagged as '%config(noreplace)' in
    the rpm rules file to make sure they are not replaced automatically if locally
    modifed.
  * Pitfals and workarounds:
    * If we're running from init, remove stale /etc/dfs/sharetab before importing
      pools in the zfs-import init script.
    * On Debian GNU/Linux, there's a 'sendsigs' script that will kill basically
      everything quite early in the shutdown phase and zed is/should be stopped
      much later than that. We don't want zed to be among the ones killed, so add
      the zed pid to list of pids for 'sendsigs' to ignore.
    * CentOS uses echo_success() and echo_failure() to print out status of
      command. These in turn uses "echo -n \0xx[etc]" to move cursor and choose
      colour etc. This doesn't work with the modified IFS variable we need to
      use in zfs-import for some reason, so work around that when we define
      zfs_log_{end,failure}_msg() for RedHat and derivative distributions.
  * All scripts passes ShellCheck (with one false positive in do_mount()).

Signed-off-by: Turbo Fredriksson turbo@bayour.com
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed by: Richard Yao <ryao@gentoo.org>
Reviewed by: Chris Dunlap <cdunlap@llnl.gov>
Closes #2974
Closes #2107
2015-05-28 14:14:53 -07:00
Brian Behlendorf
544f7184f8 Use ExecStartPre to load zfs modules
Commit 87abfcb broke the systemd import service by treating the
ExecStart line as if it were a shell command that could be executed.
This isn't the way systemd works and the correct way to handle this
case is with ExecStartPre.  This patch updates the zfs import service
files accordingly,

Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Steven Noonan <steven@uplinklabs.net>
Signed-off-by: Chris Siebenmann <cks.git01@cs.toronto.edu>
Closes #3440
2015-05-26 16:18:50 -07:00
Brian Behlendorf
65037d9b25 Add libzfs_error_init() function
All fprintf() error messages are moved out of the libzfs_init()
library function where they never belonged in the first place.  A
libzfs_error_init() function is added to provide useful error
messages for the most common causes of failure.

Additionally, in libzfs_run_process() the 'rc' variable was renamed
to 'error' for consistency with the rest of the code base.

Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Chris Dunlap <cdunlap@llnl.gov>
Signed-off-by: Richard Yao <ryao@gentoo.org>
2015-05-22 13:34:58 -07:00
Brian Behlendorf
87abfcba22 Wait in libzfs_init() for the /dev/zfs device
While module loading itself is synchronous the creation of the /dev/zfs
device is not.  This is because /dev/zfs is typically created by a udev
rule after the module is registered and presented to user space through
sysfs.  This small window between module loading and device creation
can result in spurious failures of libzfs_init().

This patch closes that race by extending libzfs_init() so it can detect
that the modules are loaded and only if required wait for the /dev/zfs
device to be created.  This allows scripts to reliably use the following
shell construct without the need for additional error handling.

$ /sbin/modprobe zfs && /sbin/zpool import -a

To minimize the potential time waiting in libzfs_init() a strategy
similar to adaptive mutexes is employed.  The function will busy-wait
for up to 10ms based on the expectation that the modules were just
loaded and therefore the /dev/zfs will be created imminently.  If it
takes longer than this it will fall back to polling for up to 10 seconds.

This behavior can be customized to some degree by setting the following
new environment variables.  This functionality is provided for backwards
compatibility with existing scripts which depend on the module auto-load
behavior.  By default module auto-loading is now disabled.

* ZFS_MODULE_LOADING="YES|yes|ON|on" - Attempt to load modules.
* ZFS_MODULE_TIMEOUT="<seconds>"     - Seconds to wait for /dev/zfs

The zfs-import-* systemd service files have been updated to call
'/sbin/modprobe zfs' so they no longer rely on the legacy auto-loading
behavior.

Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Chris Dunlap <cdunlap@llnl.gov>
Signed-off-by: Richard Yao <ryao@gentoo.org>
Closes #2556
2015-05-22 13:31:58 -07:00
Turbo Fredriksson
c11cd7f934 Style check shell scripts
If the command "shellcheck" exists, then find all shell scripts and
run shellcheck on them.
* Use 'gcc' format with shellcheck.
* Exclude zfs-script-config.sh (which isn't really a script).

Signed-off-by: Turbo Fredriksson <turbo@bayour.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #3428
2015-05-20 14:10:03 -07:00
Hajo Möller
141b6381d3 Change 3-digit octal escapes to 4-digit ones
Prefixing an octal value with a leading zero is the standard way
to disambiguate it.  This change only impacts the `zfs diff` output
and is therefore very limited in scope.

Signed-off-by: Hajo M<C3><B6>ller <dasjoe@gmail.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #3417
2015-05-18 17:14:07 -07:00
Turbo Fredriksson
01fcbec52d The mount helper mount.zfs MUST be in /sbin (not '$sbindir').
Commit 60e9f69 added the --with-mounthelperdir option for Gentoo
and in the process accidentally modified the default installation
location.  For security reasons mount(8) expects it to only be
installed under /sbin.

Signed-off-by: Turbo Fredriksson <turbo@bayour.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #3426
2015-05-18 16:54:36 -07:00
Tim Chase
f467b05a26 Initialize dbu_tqent in dmu_buf_init_user()
The dbu_evict_taskq added in 0c66c32d is only invoked via
taskq_dispatch_ent(). In these cases, ZoL's implementation of taskqs
requires the entries to be initialized first with taskq_init_ent() in
order that, among other things, the embedded spinlock is initialized
properly.

Signed-off-by: Tim Chase <tim@chase2k.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #3419
2015-05-18 11:39:14 -07:00
Alexander Eremin
7fec46b9d8 Illumos 5847 - libzfs_diff should check zfs_prop_get() return
5847 libzfs_diff should check zfs_prop_get() return
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Reviewed by: Albert Lee <trisk@omniti.com>
Approved by: Dan McDonald <danmcd@omniti.com>

References:
  https://www.illumos.org/issues/5847
  https://github.com/illumos/illumos-gate/commit/8430278

Ported-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #3412
2015-05-15 11:17:14 -07:00
Matthew Ahrens
e5fd1dd682 Illumos 5243 - zdb -b could be much faster
5243 zdb -b could be much faster
Reviewed by: Christopher Siden <christopher.siden@delphix.com>
Reviewed by: George Wilson <george.wilson@delphix.com>
Reviewed by: Richard Elling <richard.elling@gmail.com>
Approved by: Dan McDonald <danmcd@omniti.com>

References:
  https://www.illumos.org/issues/5243
  https://github.com/illumos/illumos-gate/commit/f7950bf

Ported-by: Don Brady <don.brady@intel.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #3414
2015-05-15 11:14:54 -07:00
Tim Chase
b1b85c8772 Zdb should be able to open the root dataset
If the pool/dataset command-line argument is specified with a trailing
slash, for example, "tank/", it is interpreted as the root dataset.

Signed-off-by: Tim Chase <tim@chase2k.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #3415
2015-05-15 10:52:46 -07:00
Jan Sanislo
79065ed5a4 Return -ESTALE to force lookup for missing NFS file handles
There seems to be a annoying problem using NFSv4 to access ZFS
file systems under certain circumstances.  It's easily reproduced:

    nfs_client1:  mount server:/export /mnt
    nfs_client1:  cd /mnt
    nfs_client1:  echo foo >junk
    nfs_client1:  cat junk
    foo

Now on a different NFSv4 client:

    nfs_client2:  mount server:/export /mnt
    nfs_client2:  cd /mnt
    nfs_client2:  vi junk
    # Make some changes to /mnt/junk and save
    # This change the inode associated with /mnt/junk

Now back to the original client:

    nfs_client1:  cat junk
    cat: junk: No such file or directory

Admittedly NFSv4 is not advertised as a cluster file system that
maintains a completely coherent view of data across multiple nodes.
But it does have some mechanisms built in that try to deal with
situations like the above.  Namely, it employs specialized file
handle lookup routines that return ESTALE when a file handle contains
a non-existant inode value.  The ESTALE return triggers a return
full file path lookup from the client to determine if the file has
actually gone away or if the cached file handle is no longer valid.
ZFS behavior can be brought into line with other file systems
(e.g., ext4) by applying the following patch:

Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #3224
2015-05-14 11:16:52 -07:00
Antonio Russo
7290cd3c4e Relax restriction on zfs_ioc_next_obj() iteration
Per the documentation for dnode_next_offset in dnode.c, the "txg"
parameter specifies a lower bound on which transaction the dnode can
be found in. We are interested in all dnodes that are removed between
the first and last transaction in the snapshot. It doesn't need to be
created in that snapshot to correspond to a removed file.

In fact, the behavior of zfs diff in the test case exactly matches
this: the transaction that created the data that was deleted in snapshot
"2" was produced before, in snapshot "1", definitely predating the first
transaction in snapshot "2".

Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Tim Chase <Tim Chase <tim@onlight.com>
Closes #2081
2015-05-14 11:16:08 -07:00
Brian Behlendorf
fd0fd6467b Remove unused 'dsl_pool_t *dp' variable
When ASSERTs are compiled out by using the --disable-debug configure
option.  Then the local variable 'dsl_pool_t *dp' will be unused and
generate a compiler warning.  Since this variable is only used once
in the ASSERT replace it with 'ds->ds_dir->dd_pool'.

This has the additional advantage of potentially saving a few bytes
on the stack depending on how gcc decides to compile the function.

This issue was not noticed immediately because the automated builders
use --enable-debug to make the testing as rigorous as possible.

Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Chris Dunlap <cdunlap@llnl.gov>
Closes #3410
2015-05-14 11:09:47 -07:00
Max Grossman
5dc8b7365f Illumos 5765 - add support for estimating send stream size with lzc_send_space when source is a bookmark
5765 add support for estimating send stream size with lzc_send_space when source is a bookmark
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Reviewed by: Christopher Siden <christopher.siden@delphix.com>
Reviewed by: Steven Hartland <killing@multiplay.co.uk>
Reviewed by: Bayard Bell <buffer.g.overflow@gmail.com>
Approved by: Albert Lee <trisk@nexenta.com>

References:
  https://www.illumos.org/issues/5765
  https://github.com/illumos/illumos-gate/commit/643da460

Porting notes:
* Unused variable 'recordsize' in dmu_send_estimate() dropped

Ported-by: DHE <git@dehacked.net>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #3397
2015-05-13 09:03:59 -07:00
Justin T. Gibbs
19b3b1d2a2 Illumos 5393 - spurious failures from dsl_dataset_hold_obj()
5393 spurious failures from dsl_dataset_hold_obj()
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Reviewed by: Will Andrews <willa@spectralogic.com>
Reviewed by: Prakash Surya <prakash.surya@delphix.com>
Reviewed by: Steven Hartland <killing@multiplay.co.uk>
Approved by: Dan McDonald <danmcd@omniti.com>

References:
  https://www.illumos.org/issues/5393
  https://github.com/illumos/illumos-gate/commit/e1f3c20

Ported-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #3403
2015-05-13 08:56:04 -07:00
Justin T. Gibbs
63b33e878a Illumos 5562 - ZFS sa_handle's violate kmem invariants, debug kernels panic on boot
5562 ZFS sa_handle's violate kmem invariants, debug kernels panic on boot
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Reviewed by: Robert Mustacchi <rm@fingolfin.org>
Reviewed by: George Wilson <george@delphix.com>
Reviewed by: Rich Lowe <richlowe@richlowe.net>
Approved by: Dan McDonald <danmcd@omniti.com>

References:
  https://www.illumos.org/issues/5562
  https://github.com/illumos/illumos-gate/commit/0fda3cc5

Ported-by: DHE <git@dehacked.net>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #3388
2015-05-11 15:10:57 -07:00
Matthew Ahrens
252e1a54ab Illumos 5810 - zdb should print details of bpobj
5810 zdb should print details of bpobj
Reviewed by: Prakash Surya <prakash.surya@delphix.com>
Reviewed by: Alex Reece <alex@delphix.com>
Reviewed by: George Wilson <george@delphix.com>
Reviewed by: Will Andrews <will@freebsd.org>
Reviewed by: Simon Klinkert <simon.klinkert@gmail.com>
Approved by: Gordon Ross <gwr@nexenta.com>

References:
  https://www.illumos.org/issues/5810
  https://github.com/illumos/illumos-gate/commit/732885fc

Ported-by: DHE <git@dehacked.net>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #3387
2015-05-11 15:10:24 -07:00