We should use zfs_dbgmsg instead of spa_dbgmsg. Or at least,
metaslab_condense() should call zfs_dbgmsg because it's important and rare
enough to always log. It's possible that the message in zio_dva_allocate()
would be too high-frequency for zfs_dbgmsg.
illumos/illumos-gate@21f7c81cc1
Reviewed by: Serapheim Dimitropoulos <serapheim.dimitro@delphix.com>
Reviewed by: Pavel Zakharov <pavel.zakharov@delphix.com>
Reviewed by: George Wilson <george.wilson@delphix.com>
Reviewed by: Richard Elling <Richard.Elling@RichardElling.com>
Approved by: Richard Lowe <richlowe@richlowe.net>
Author: Matthew Ahrens <mahrens@delphix.com>
Currently vdev_label_sync and vdev_uberblock_sync take a zio_t and assume
that its io_private is a pointer to the good_writes count. They should
instead accept this argument explicitly.
illumos/illumos-gate@a3b5583021
Reviewed by: Pavel Zakharov <pavel.zakharov@delphix.com>
Reviewed by: George Wilson <george.wilson@delphix.com>
Approved by: Richard Lowe <richlowe@richlowe.net>
Author: Matthew Ahrens <mahrens@delphix.com>
Mirrors are supposed to provide redundancy in the face of whole-disk failure
and silent damage (e.g. some data on disk is not right, but ZFS hasn't
detected the whole device as being broken). However, the current device
removal implementation bypasses some of the mirror's redundancy.
illumos/illumos-gate@3a4b1be953
Reviewed by: George Wilson <george.wilson@delphix.com>
Reviewed by: Prashanth Sreenivasa <pks@delphix.com>
Reviewed by: Sara Hartse <sara.hartse@delphix.com>
Reviewed by: Serapheim Dimitropoulos <serapheim@delphix.com>
Reviewed by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed by: Tim Chase <tim@chase2k.com>
Approved by: Richard Lowe <richlowe@richlowe.net>
Author: Matthew Ahrens <mahrens@delphix.com>
On high-end systems running async sequential write workloads, especially
NUMA systems with flash or NVMe storage, one significant performance
bottleneck is selecting a metaslab to do allocations from. This process
can be parallelized, providing significant performance increases for
these workloads.
illumos/illumos-gate@f78cdc34af
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Reviewed by: George Wilson <george.wilson@delphix.com>
Reviewed by: Serapheim Dimitropoulos <serapheim.dimitro@delphix.com>
Reviewed by: Alexander Motin <mav@FreeBSD.org>
Approved by: Gordon Ross <gwr@nexenta.com>
Author: Paul Dagnelie <pcd@delphix.com>
The current space map encoding has the following disadvantages:
[1] Assuming 512 sector size each entry can represent at most 16MB for a segment.
This makes the encoding very inefficient for large regions of space.
[2] As vdev-wide space maps have started to be used by new features (i.e.
device removal, zpool checkpoint) we've started imposing limits in the
vdevs that can be used with them based on the maximum addressable offset
(currently 64PB for a top-level vdev).
The new remains backwards compatible with the old one. The introduced
two-word entry format, besides extending the limits imposed by the single-entry
layout, also includes a vdev field and some extra padding after its prefix.
The extra padding after the prefix should is reserved for future usage (e.g.
new prefixes for future encodings or new fields for flags). The new vdev field
not only makes the space maps more self-descriptive, but also opens the doors
for pool-wide space maps.
One final important note is that the number of bits used for vdevs is reduced
to 24 bits for blkptrs. That was decided as we don't know of any setups that
use more than 16M vdevs for the time being and
we wanted to fit the vdev field in the space map. In addition that gives us
some extra bits in dva_t.
illumos/illumos-gate@17f11284b4
Reviewed by: Matt Ahrens <mahrens@delphix.com>
Reviewed by: George Wilson <gwilson@zfsmail.com>
Approved by: Gordon Ross <gwr@nexenta.com>
Author: Serapheim Dimitropoulos <serapheim@delphix.com>
illumos/illumos-gate@b6bf6e1540
Reviewed by: George Wilson <george.wilson@delphix.com>
Reviewed by: Prashanth Sreenivasa <pks@delphix.com>
Approved by: Matt Ahrens <mahrens@delphix.com>
Author: Pavel Zakharov <pavel.zakharov@delphix.com>
`arc_reclaim_thread()` calls `arc_adjust()` after calling
`arc_kmem_reap_now()`; `arc_adjust()` signals `arc_get_data_buf()` to
indicate that we may no longer be `arc_is_overflowing()`.
The problem is, `arc_kmem_reap_now()` can take several seconds to
complete, has no impact on `arc_is_overflowing()`, but due to how the
code is structured, can impact how long the ARC will remain in the
`arc_is_overflowing()` state.
The fix is to use seperate threads to:
1. keep `arc_size` under `arc_c`, by calling `arc_adjust()`, which
improves `arc_is_overflowing()`
2. keep enough free memory in the system, by calling
`arc_kmem_reap_now()` plus `arc_shrink()`, which improves
`arc_available_memory()`.
illumos/illumos-gate@de753e34f9
Reviewed by: Matt Ahrens <mahrens@delphix.com>
Reviewed by: Serapheim Dimitropoulos <serapheim@delphix.com>
Reviewed by: Pavel Zakharov <pavel.zakharov@delphix.com>
Reviewed by: Dan Kimmel <dan.kimmel@delphix.com>
Reviewed by: Paul Dagnelie <pcd@delphix.com>
Reviewed by: Dan McDonald <danmcd@joyent.com>
Reviewed by: Tim Kordas <tim.kordas@joyent.com>
Approved by: Garrett D'Amore <garrett@damore.org>
Author: Brad Lewis <brad.lewis@delphix.com>
illumos/illumos-gate@243952c7ee
Reviewed by: George Wilson <george.wilson@delphix.com>
Reviewed by: John Kennedy <john.kennedy@delphix.com>
Approved by: Garrett D'Amore <garrett@damore.org>
Author: Matt Ahrens <Matt.Ahrens@delphix.com>
illumos/illumos-gate@268bbb2a2f
With compressed ARC (6950) we use up to 25% of our CPU to decompress indirect
blocks, under a workload of random cached reads. To reduce this decompression
cost, we would like to increase the size of the dbuf cache so that more
indirect blocks can be stored uncompressed.
If we are caching entire large files of recordsize=8K, the indirect blocks
use 1/64th as much memory as the data blocks (assuming they have the same
compression ratio). We suggest making the dbuf cache be 1/32nd of all memory,
so that in this scenario we should be able to keep all the indirect blocks
decompressed in the dbuf cache. (We want it to be more than the 1/64th that
the indirect blocks would use because we need to cache other stuff in the
dbuf cache as well.)
In real world workloads, this won't help as dramatically as the example
above, but we think it's still worth it because the risk of decreasing
performance is low. The potential negative performance impact is that we
will be slightly reducing the size of the ARC (by ~3%).
Reviewed by: Dan Kimmel <dan.kimmel@delphix.com>
Reviewed by: Prashanth Sreenivasa <pks@delphix.com>
Reviewed by: Paul Dagnelie <pcd@delphix.com>
Reviewed by: Sanjay Nadkarni <sanjay.nadkarni@nexenta.com>
Reviewed by: Allan Jude <allanjude@freebsd.org>
Reviewed by: Igor Kozhukhov <igor@dilos.org>
Approved by: Garrett D'Amore <garrett@damore.org>
Author: George Wilson <george.wilson@delphix.com>
illumos/illumos-gate@9be12bd737
arc_loan_compressed_buf() increments arc_loaned_bytes by psize unconditionally
In the case of zfs_compressed_arc_enabled=0, when the buf is returned via
arc_return_buf(), if ARC_BUF_COMPRESSED(buf) is false, then arc_loaned_bytes
is decremented by lsize, not psize.
Switch to using arc_buf_size(buf), instead of psize, which will return
psize or lsize, depending on the result of ARC_BUF_COMPRESSED(buf).
Reviewed by: Matt Ahrens <matt@delphix.com>
Reviewed by: George Wilson <george.wilson@delphix.com>
Approved by: Garrett D'Amore <garrett@damore.org>
Author: Allan Jude <allanjude@freebsd.org>
illumos/illumos-gate@5dafeea3eb
We want to be able to pass various settings during import/open of a pool,
which are not only related to rewind. Instead of adding a new policy and
duplicate a bunch of code, we should just rename rewind_policy to a more
generic term like load_policy.
For instance, we'd like to set spa->spa_import_flags from the nvlist,
rather from a flags parameter passed to spa_import as in some cases we want
those flags not only for the import case, but also for the open case. One
such flag could be ZFS_IMPORT_MISSING_LOG (as used in zdb) which would
allow zfs to open a pool when logs are missing.
Reviewed by: Matt Ahrens <matt@delphix.com>
Reviewed by: George Wilson <george.wilson@delphix.com>
Approved by: Robert Mustacchi <rm@joyent.com>
Author: Pavel Zakharov <pavel.zakharov@delphix.com>
illumos/illumos-gate@ccef24b493
Reviewed by: George Wilson <george.wilson@delphix.com>
Reviewed by: Prakash Surya <prakash.surya@delphix.com>
Reviewed by: Matt Ahrens <matt@delphix.com>
Approved by: Robert Mustacchi <rm@joyent.com>
Author: Pavel Zakharov <pavel.zakharov@delphix.com>
illumos/illumos-gate@d1de72cfa2
ztest failed with uncorrectable IO error despite having the fix for #7163.
Both sides of the mirror have CANT_OPEN_BAD_LABEL, which also distinguishes
it from that issue.
Definitely seems like a racing condition between the vdev_validate and spa_sync:
1. Thread A (spa_sync): vdev label is updated to latest txg
2. Thread B (vdev_validate): vdev label's txg is compared to spa_last_synced_txg and is ahead.
3. Thread A (spa_sync): spa_last_synced_txg is updated to latest txg.
Solution: do not check txg in vdev_validate unless config lock is held.
Reviewed by: George Wilson <george.wilson@delphix.com>
Reviewed by: Matt Ahrens <matthew.ahrens@delphix.com>
Approved by: Robert Mustacchi <rm@joyent.com>
Author: Pavel Zakharov <pavel.zakharov@delphix.com>
illumos/illumos-gate@8671400134
The idea of Storage Pool Checkpoint (aka zpool checkpoint) deals with
exactly that. It can be thought of as a “pool-wide snapshot” (or a
variation of extreme rewind that doesn’t corrupt your data). It remembers
the entire state of the pool at the point that it was taken and the user
can revert back to it later or discard it. Its generic use case is an
administrator that is about to perform a set of destructive actions to ZFS
as part of a critical procedure. She takes a checkpoint of the pool before
performing the actions, then rewinds back to it if one of them fails or puts
the pool into an unexpected state. Otherwise, she discards it. With the
assumption that no one else is making modifications to ZFS, she basically
wraps all these actions into a “high-level transaction”.
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Reviewed by: John Kennedy <john.kennedy@delphix.com>
Reviewed by: Dan Kimmel <dan.kimmel@delphix.com>
Approved by: Richard Lowe <richlowe@richlowe.net>
Author: Serapheim Dimitropoulos <serapheim.dimitro@delphix.com>
illumos/illumos-gate@edc8ef7d92
Reviewed by: C Fraire <cfraire@me.com>
Reviewed by: Andy Fiddaman <omnios@citrus-it.co.uk>
Approved by: Joshua M. Clulow <josh@sysmgr.org>
Author: Toomas Soome <tsoome@me.com>
In pursuit of improving performance on multi-core systems, we should
implements fanned out counters and use them to improve the performance of
some of the arc statistics. These stats are updated extremely frequently,
and can consume a significant amount of CPU time.
Reviewed by: Pavel Zakharov <pavel.zakharov@delphix.com>
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Approved by: Dan McDonald <danmcd@joyent.com>
Author: Paul Dagnelie <pcd@delphix.com>
illumos/illumos-gate@5f5913bb835f5913bb83https://www.illumos.org/issues/9164
This issue has been reported by Alan Somers as
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=225877
dmu_objset_refresh_ownership() first disowns a dataset (and releases
it) and then owns it again. There is an assert that the new dataset
object is the same as the old dataset object. When running ZFS Test
Suite on FreeBSD we see this panic from zpool_upgrade_007_pos test:
panic: solaris assert: newds == os->os_dsl_dataset (0xfffff80045f4c000
== 0xfffff80021ab4800)
I see that the old dataset has dsl_dataset_evict_async() pending in
ds_dbu.dbu_tqent and its ds_dbuf is NULL.
Reviewed by: Matt Ahrens <mahrens@delphix.com>
Reviewed by: Don Brady <don.brady@delphix.com>
Approved by: Richard Lowe <richlowe@richlowe.net>
Author: Andriy Gapon <avg@FreeBSD.org>
illumos/illumos-gate@e9bacc6d1ae9bacc6d1ahttps://www.illumos.org/issues/8984
Consider a directory configured as:
drwx-ws---+ 2 henson cpp 3 Jan 23 12:35 dropbox/
user:henson:rwxpdDaARWcC--:f-i----:allow
owner@:--------------:f-i----:allow
group@:--------------:f-i----:allow
everyone@:--------------:f-i----:allow
owner@:rwxpdDaARWcC--:-di----:allow
group:cpp:-wx-----------:-------:allow
owner@:rwxpdDaARWcC--:-------:allow
A new file created in this directory ends up looking like:
rw-r--r-+ 1 astudent cpp 0 Jan 23 12:39 testfile
user:henson:rw-pdDaARWcC--:------I:allow
owner@:--------------:------I:allow
group@:--------------:------I:allow
everyone@:--------------:------I:allow
owner@:rw-p--aARWcCos:-------:allow
group@:r-----a-R-c--s:-------:allow
everyone@:r-----a-R-c--s:-------:allow
with extraneous group@ and everyone@ entries allowing read access that
shouldn't exist.
Per Albert Lee on the zfs mailing list:
"aclinherit=passthrough/passthrough-x should still
ignore the requested mode when an inheritable ACE for owner@ group@,
or everyone@ is present in the parent directory.
It appears there was an oversight in my fix for
https://www.illumos.org/issues/6764 which made calling zfs_acl_chmod
from zfs_acl_inherit unconditional. I think the parent ACL check for
aclinherit=passthrough needs to be reintroduced in zfs_acl_inherit."
We have a large number of faculty who use dropbox directories like the example
to have students submit projects. All of these directories are now allowing
Reviewed by: Sam Zaydel <szaydel@racktopsystems.com>
Reviewed by: Paul B. Henson <henson@acm.org>
Reviewed by: Prakash Surya <prakash.surya@delphix.com>
Approved by: Matthew Ahrens <mahrens@delphix.com>
Author: Dominik Hassler <hadfl@omniosce.org>
illumos/illumos-gate@bdfded42e6
A scenario came up where a callback executed by vdev_indirect_remap() on a vdev, calls
vdev_indirect_remap() on the same vdev and tries to reacquire vdev_indirect_rwlock that
was already acquired from the first call to vdev_indirect_remap(). The specific scenario,
is that we want to remap a block pointer that is snapshoted but its dataset's remap_deadlist
is not cached. So in order to add it we issue a read through a vdev_indirect_remap() on the
same vdev, which brings up the aforementioned issue.
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Reviewed by: George Wilson <george.wilson@delphix.com>
Approved by: Hans Rosenfeld <rosenfeld@grumpf.hope-2000.org>
Author: Serapheim Dimitropoulos <serapheim.dimitro@delphix.com>
illumos/illumos-gate@667ec66f1b
The timeline of the race condition is the following:
[1] Thread A is about to finish condesing the first vdev in spa_condense_indirect_thread(),
so it calls the spa_condense_indirect_complete_sync() sync task which sets the
spa_condensing_indirect field to NULL. Waiting for the sync task to finish, thread A
sleeps until the txg is done. When this happens, thread A will acquire spa_async_lock
and set spa_condense_thread to NULL.
[2] While thread A waits for the txg to finish, thread B which is running spa_sync() checks
whether it should condense the second vdev in vdev_indirect_should_condense() by checking
the spa_condensing_indirect field which was set to NULL by spa_condense_indirect_thread()
from thread A. So it goes on and tries to spawn a new condensing thread in
spa_condense_indirect_start_sync() and the aforementioned assertions fails because thread A
has not set spa_condense_thread to NULL (which is basically the last thing it does before
returning).
The main issue here is that we rely on both spa_condensing_indirect and spa_condense_thread to
signify whether a condensing thread is running. Ideally we would only use one throughout the
codebase. In addition, for managing spa_condense_thread we currently use spa_async_lock which
basically tights condensing to scrubing when it comes to pausing and resuming those actions
during spa export.
Reviewed by: Matt Ahrens <mahrens@delphix.com>
Reviewed by: Pavel Zakharov <pavel.zakharov@delphix.com>
Approved by: Hans Rosenfeld <rosenfeld@grumpf.hope-2000.org>
Author: Serapheim Dimitropoulos <serapheim@delphix.com>
9075 Improve ZFS pool import/load process and corrupted pool recovery
illumos/illumos-gate@6f7938128a
Some work has been done lately to improve the debugability of the ZFS pool
load (and import) process. This includes:
https://www.illumos.org/issues/7638: Refactor spa_load_impl into several functions
https://www.illumos.org/issues/8961: SPA load/import should tell us why it failed
https://www.illumos.org/issues/7277: zdb should be able to print zfs_dbgmsg's
To iterate on top of that, there's a few changes that were made to make the
import process more resilient and crash free. One of the first tasks during the
pool load process is to parse a config provided from userland that describes
what devices the pool is composed of. A vdev tree is generated from that config,
and then all the vdevs are opened.
The Meta Object Set (MOS) of the pool is accessed, and several metadata objects
that are necessary to load the pool are read. The exact configuration of the
pool is also stored inside the MOS. Since the configuration provided from
userland is external and might not accurately describe the vdev tree
of the pool at the txg that is being loaded, it cannot be relied upon to safely
operate the pool. For that reason, the configuration in the MOS is read early
on. In the past, the two configurations were compared together and if there was
a mismatch then the load process was aborted and an error was returned.
The latter was a good way to ensure a pool does not get corrupted, however it
made the pool load process needlessly fragile in cases where the vdev
configuration changed or the userland configuration was outdated. Since the MOS
is stored in 3 copies, the configuration provided by userland doesn't have to be
perfect in order to read its contents. Hence, a new approach has been adopted:
The pool is first opened with the untrusted userland configuration just so that
the real configuration can be read from the MOS. The trusted MOS configuration
is then used to generate a new vdev tree and the pool is re-opened.
When the pool is opened with an untrusted configuration, writes are disabled
to avoid accidentally damaging it. During reads, some sanity checks are
performed on block pointers to see if each DVA points to a known vdev;
when the configuration is untrusted, instead of panicking the system if those
checks fail we simply avoid issuing reads to the invalid DVAs.
This new two-step pool load process now allows rewinding pools accross
vdev tree changes such as device replacement, addition, etc. Loading a pool
from an external config file in a clustering environment also becomes much
safer now since the pool will import even if the config is outdated and didn't,
for instance, register a recent device addition.
With this code in place, it became relatively easy to implement a
long-sought-after feature: the ability to import a pool with missing top level
(i.e. non-redundant) devices. Note that since this almost guarantees some loss
Of data, this feature is for now restricted to a read-only import.
Reviewed by: George Wilson <george.wilson@delphix.com>
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Reviewed by: Andrew Stormont <andyjstormont@gmail.com>
Approved by: Hans Rosenfeld <rosenfeld@grumpf.hope-2000.org>
Author: Pavel Zakharov <pavel.zakharov@delphix.com>
illumos/illumos-gate@add927f8c8
Reported on the ZFSonLinux https://github.com/zfsonlinux/zfs/issues/4843,
fixed by https://github.com/zfsonlinux/zfs/pull/6339:
If we are in the middle of an incremental zfs receive, the child .../%recv
will exist. If you concurrently run zfs promote .../%recv, it will "work",
but then zfs gets confused. For example, there's no obvious way to destroy
the containing filesystem (because it is now a clone of its invisible child).
Attempting to do this promote should be an error. We could fix this by
having zfs_ioc_promote() check if zc_name contains a %, similar to
zfs_ioc_rename().
Reviewed by: Paul Dagnelie <pcd@delphix.com>
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Approved by: Dan McDonald <danmcd@joyent.com>
Author: loli10K <ezomori.nozomu@gmail.com>
illumos/illumos-gate@f4c1745bd6
Illumos 4080 allows "zpool clear" to work on readonly pools: i don't think
this is the intended behaviour, we shouldn't be allowed to clear readonly
pools. Probably.
A fix is already in the ZFS on Linux repository to addess this issue:
92e43c1718
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Approved by: Dan McDonald <danmcd@joyent.com>
Author: loli10K <ezomori.nozomu@gmail.com>
illumos/illumos-gate@85723e5eec
When iterating over the input nvlist in dsl_props_set_sync_impl() when we
don't preserve the nvpair name before looking up ZPROP_VALUE, so when we
later go to process it nvpair_name() is always "value" instead of the actual
property name.
This results in a couple of bugs in the recv code:
- received properties are not restored correctly when failing to receive
an incremental send stream
- received properties are not completely replaced by the new ones when
successfully receiving an incremental send stream
This was discovered on ZFS on Linux (fixed in
5f1346c299)
Reviewed by: Paul Dagnelie <pcd@delphix.com>
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Approved by: Dan McDonald <danmcd@joyent.com>
Author: loli10K <ezomori.nozomu@gmail.com>
illumos/illumos-gate@f02c28e434
Reviewed by: Yuri Pankov <yuripv@yuripv.net>
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Approved by: Dan McDonald <danmcd@joyent.com>
Author: Toomas Soome <tsoome@me.com>
illumos/illumos-gate@46ac8fdfc5
Reviewed by: Yuri Pankov <yuripv@yuripv.net>
Reviewed by: Andy Fiddaman <omnios@citrus-it.co.uk>
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Approved by: Dan McDonald <danmcd@joyent.com>
Author: Toomas Soome <tsoome@me.com>
illumos/illumos-gate@e144c4e6c9
Currently `zdb` consistently fails to examine non-idle pools as it fails
during the `spa_load()` process. The main problem seems to be that
`spa_load_verify()` fails as can be seen below:
$ sudo zdb -d -G dcenter
zdb: can't open 'dcenter': I/O error
ZFS_DBGMSG(zdb):
spa_open_common: opening dcenter
spa_load(dcenter): LOADING
disk vdev '/dev/dsk/c4t11d0s0': best uberblock found for spa dcenter. txg 40824950
spa_load(dcenter): using uberblock with txg=40824950
spa_load(dcenter): UNLOADING
spa_load(dcenter): RELOADING
spa_load(dcenter): LOADING
disk vdev '/dev/dsk/c3t10d0s0': best uberblock found for spa dcenter. txg 40824952
spa_load(dcenter): using uberblock with txg=40824952
spa_load(dcenter): FAILED: spa_load_verify failed [error=5]
spa_load(dcenter): UNLOADING
This change makes `spa_load_verify()` a dryrun when ran from `zdb`. This is
done by creating a global flag in zfs and then setting it in `zdb`.
Reviewed by: George Wilson <george.wilson@delphix.com>
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Reviewed by: Andy Stormont <astormont@racktopsystems.com>
Approved by: Dan McDonald <danmcd@joyent.com>
Author: Pavel Zakharov <pavel.zakharov@delphix.com>
illumos/illumos-gate@3ee8c80c74
When we fail to open or import a storage pool, we typically don't get any
additional diagnostic information, just "no pool found" or "can not import".
While there may be no additional user-consumable information, we should at
least make this situation easier to debug/diagnose for developers and support.
For example, we could start by using `zfs_dbgmsg()` to log each thing that we
try when importing, and which things failed. E.g. "tried uberblock of txg X
from label Y of device Z". Also, we could log each of the stages that we go
through in `spa_load_impl()`.
Reviewed by: George Wilson <george.wilson@delphix.com>
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Reviewed by: Andrew Stormont <andyjstormont@gmail.com>
Approved by: Dan McDonald <danmcd@joyent.com>
Author: Pavel Zakharov <pavel.zakharov@delphix.com>
illumos/illumos-gate@1fd3785ff6
spa_load_impl has grown out of proportions. It is currently over 700
lines long and makes it very hard to follow or debug the import process
even for experienced ZFS developers. The objective is to split it up
in a series of well commented functions.
Reviewed by: Paul Dagnelie <pcd@delphix.com>
Reviewed by: George Wilson <george.wilson@delphix.com>
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Reviewed by: Andrew Stormont <andyjstormont@gmail.com>
Approved by: Dan McDonald <danmcd@joyent.com>
Author: Pavel Zakharov <pavel.zakharov@delphix.com>
illumos/illumos-gate@36a64e6284
To prevent kmem_cache reaping from blocking other system resources, turn
kmem_cache_reap_now() (which blocks) into kmem_cache_reap_soon(). Callers
to kmem_cache_reap_soon() should use kmem_cache_reap_active(), which
exploits #9017's new taskq_empty().
Reviewed by: Bryan Cantrill <bryan@joyent.com>
Reviewed by: Dan McDonald <danmcd@joyent.com>
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Reviewed by: Yuri Pankov <yuripv@yuripv.net>
Author: Tim Kordas <tim.kordas@joyent.com>
illumos/illumos-gate@f06dce2c1f
Reviewed by: Sebastien Roy <sebastien.roy@delphix.com>
Reviewed by: Prakash Surya <prakash.surya@delphix.com>
Reviewed by: Gordon Ross <gordon.w.ross@gmail.com>
Approved by: Richard Lowe <richlowe@richlowe.net>
Author: Andrew Stormont <astormont@racktopsystems.com>
illumos/illumos-gate@0fb055e81f
At present it is possible to boot from a root pool that is on RAIDZ but not
one that is on RAIDZ2 or RAIDZ3. This is because, at the time the pool
version is checked to ensure support for dual/triple parity, the uberblock
has not yet been loaded into the SPA and therefore the code determines that
the pool version is too old and returns ENOTSUP.
Reviewed by: Igor Kozhukhov <igor@dilos.org>
Reviewed by: Andriy Gapon <avg@FreeBSD.org>
Reviewed by: Pavel Zakharov <pavel.zakharov@delphix.com>
Reviewed by: Andy Stormont <astormont@racktopsystems.com>
Reviewed by: Toomas Soome <tsoome@me.com>
Approved by: Gordon Ross <gwr@nexenta.com>
Author: Andy Fiddaman <omnios@citrus-it.co.uk>
illumos/illumos-gate@95643f75d295643f75d2https://www.illumos.org/issues/8520
lzc_rollback_to() should support rolling back to a clone's origin.
The current checks in zfs_ioc_rollback() would not allow that because the
origin snapshot belongs to a different filesystem.
The overly restrictive check was introduced in 7600, but it was not a
regression as none of the existing tools provided a way to rollback to the
origin.
https://www.illumos.org/issues/7198
EINVAL is returned when a dataset does not have any snapshots, so there is
nothing to roll back to.
Although the code in zfs_do_rollback checks for that condition in advance, it's
still possible that the snapshot(s) gets removed after the check and before the
rollback sync task is executed.
At the moment zfs command would crash when that happens.
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Approved by: Dan McDonald <danmcd@joyent.com>
Author: Andriy Gapon <avg@FreeBSD.org>
illumos/illumos-gate@f864f99efef864f99efehttps://www.illumos.org/issues/8997
When dmu_tx_assign is called from zil_lwb_write_issue, it's possible
for either ERESTART or EIO to be returned.
If ERESTART is returned, this will cause an assertion to fail directly
in zil_lwb_write_issue, where the code assumes the return value is
EIO if dmu_tx_assign returns a non-zero value. This can occur if the
SPA is suspended when dmu_tx_assign is called, and most often occurs
when running zloop.
If EIO is returned, this can cause assertions to fail elsewhere in the
ZIL code. For example, zil_commit_waiter_timeout contains the
following logic:
lwb_t *nlwb = zil_lwb_write_issue(zilog, lwb);
ASSERT3S(lwb->lwb_state, !=, LWB_STATE_OPENED);
In this case, if dmu_tx_assign returned EIO from within
zil_lwb_write_issue, the lwb variable passed in will not be issued
to disk. Thus, it's lwb_state field will remain LWB_STATE_OPENED and
this assertion will fail. zil_commit_waiter_timeout assumes that after
it calls zil_lwb_write_issue, the lwb will be issued to disk, and
doesn't handle the case where this is not true; i.e. it doesn't handle
the case where dmu_tx_assign returns EIO.
Reviewed by: Matt Ahrens <mahrens@delphix.com>
Reviewed by: Andriy Gapon <avg@FreeBSD.org>
Approved by: Robert Mustacchi <rm@joyent.com>
Author: Prakash Surya <prakash.surya@delphix.com>
illumos/illumos-gate@a6c1eb3c08a6c1eb3c08https://www.illumos.org/issues/8731
annotate_ecksum() asserts that nui64s, calculated as nui64s = size / sizeof
(uint64_t), is not greater than UINT16_MAX.
This restriction is needed because histograms of incorrectly set and cleared
bits have 16 bit counters and if the buffer consists of too many 64-bit words,
then a counter can potentially overflow producing an incorrect result.
When the largest buffer size was 128KB the greatest value of nui64s was 16K,
well within the limit.
But now we have support for large buffers and for buffer sizes of 512KB and
above the restriction is violated.
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Approved by: Dan McDonald <danmcd@joyent.com>
Author: Andriy Gapon <avg@FreeBSD.org>
illumos/illumos-gate@5cabbc6b49https://www.illumos.org/issues/7614:
This project allows top-level vdevs to be removed from the storage pool with
“zpool remove”, reducing the total amount of storage in the pool. This
operation copies all allocated regions of the device to be removed onto other
devices, recording the mapping from old to new location. After the removal is
complete, read and free operations to the removed (now “indirect”) vdev must
be remapped and performed at the new location on disk. The indirect mapping
table is kept in memory whenever the pool is loaded, so there is minimal
performance overhead when doing operations on the indirect vdev.
The size of the in-memory mapping table will be reduced when its entries
become “obsolete” because they are no longer used by any block pointers in
the pool. An entry becomes obsolete when all the blocks that use it are
freed. An entry can also become obsolete when all the snapshots that
reference it are deleted, and the block pointers that reference it have been
“remapped” in all filesystems/zvols (and clones). Whenever an indirect block
is written, all the block pointers in it will be “remapped” to their new
(concrete) locations if possible. This process can be accelerated by using
the “zfs remap” command to proactively rewrite all indirect blocks that
reference indirect (removed) vdevs.
Note that when a device is removed, we do not verify the checksum of the data
that is copied. This makes the process much faster, but if it were used on
redundant vdevs (i.e. mirror or raidz vdevs), it would be possible to copy
the wrong data, when we have the correct data on e.g. the other side of the
mirror. Therefore, mirror and raidz devices can not be removed.
Reviewed by: Alex Reece <alex@delphix.com>
Reviewed by: George Wilson <george.wilson@delphix.com>
Reviewed by: John Kennedy <john.kennedy@delphix.com>
Reviewed by: Prakash Surya <prakash.surya@delphix.com>
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Reviewed by: Richard Laager <rlaager@wiktel.com>
Reviewed by: Tim Chase <tim@chase2k.com>
Approved by: Garrett D'Amore <garrett@damore.org>
Author: Prashanth Sreenivasa <pks@delphix.com>
illumos/illumos-gate@d6e1c446d7d6e1c446d7https://www.illumos.org/issues/8857
I had an OS panic on one of our servers:
ffffff01809128c0 vpanic()
ffffff01809128e0 mutex_panic+0x58(fffffffffb94c904, ffffff597dde7f80)
ffffff0180912950 mutex_vector_enter+0x347(ffffff597dde7f80)
ffffff01809129b0 zio_remove_child+0x50(ffffff597dde7c58, ffffff32bd901ac0,
ffffff3373370908)
ffffff0180912a40 zio_done+0x390(ffffff32bd901ac0)
ffffff0180912a70 zio_execute+0x78(ffffff32bd901ac0)
ffffff0180912b30 taskq_thread+0x2d0(ffffff33bae44140)
ffffff0180912b40 thread_start+8()
It panicked here:
http://src.illumos.org/source/xref/illumos-gate/usr/src/uts/common/fs/zfs/
zio.c#430
pio->io_lock is DEAD, thus a panic. Further analysis shows the "pio"
(parent zio of "cio") has already been destroyed.
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Reviewed by: Andriy Gapon <avg@FreeBSD.org>
Reviewed by: Youzhong Yang <youzhong@gmail.com>
Approved by: Dan McDonald <danmcd@omniti.com>
Author: George Wilson <george.wilson@delphix.com>
illumos/illumos-gate@5cb8d943bchttps://www.illumos.org/issues/8835:
Sequential reads not aligned to block size are not detected by ZFS
prefetcher as sequential, killing prefetch and severely hurting
performance. It is caused by dmu_zfetch() in case of misaligned
sequential accesses being called with overlap of one block.
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Reviewed by: Allan Jude <allanjude@freebsd.org>
Approved by: Gordon Ross <gwr@nexenta.com>
Author: Alexander Motin <mav@FreeBSD.org>
illumos/illumos-gate@4ae5f5f06chttps://www.illumos.org/issues/8652:
Clang and GCC prefer to use unsigned ints to store enums. With Clang, that
causes tautological comparison warnings when comparing a zfs_prop_t or
zpool_prop_t variable to the macro ZPROP_INVAL. It's likely that error
handling code is being silently removed as a result.
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Reviewed by: Igor Kozhukhov <igor@dilos.org>
Approved by: Gordon Ross <gwr@nexenta.com>
Author: Alan Somers <asomers@gmail.com>
illumos/illumos-gate@301fd1d6f2
Reviewed by: Alek Pinchuk <pinchuk.alek@gmail.com>
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Approved by: Gordon Ross <gwr@nexenta.com>
Author: Sean Eric Fagan <sef@ixsystems.com>
illumos/illumos-gate@01a059ee0chttps://www.illumos.org/issues/8856:
arc_cksum_is_equal() calls zio_push_transform() that requires abd_t*
(second arg), but a void* is passed.
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Reviewed by: George Wilson <george.wilson@delphix.com>
Approved by: Gordon Ross <gwr@nexenta.com>
Author: Roman Strashkin <roman.strashkin@nexenta.com>
illumos/illumos-gate@93c618e0f4https://www.illumos.org/issues/8930:
We normally remove an unlinked node when its last user goes away and the
node becomes inactive. However, we should not do that if the filesystem
is mounted read-only including the case where it has its readonly
property set. The node will remain on the unlinked queue, so it will
not be leaked.
One particular scenario is when we receive an incremental stream into a
mounted read-only filesystem and that stream contains an unlinked file
(still on the unlinked queue). If that file is opened before the
receive and some time later after the receive it becomes inactive we
would remove it and, thus, modify the read-only filesystem. As a
result, the filesystem would diverge from its source and further
incremental receives would not be possible (without forcing a rollback).
Another related scenario, that may or may not be possible depending on an
OS / VFS policy, is when an open file is unlinked, then the filesystem is
remounted read-only, and then the file is closed.
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Approved by: Gordon Ross <gwr@nexenta.com>
Author: Andriy Gapon <avg@FreeBSD.org>
illumos/illumos-gate@94ddd0900ahttps://www.illumos.org/issues/8909:
There's a race condition that exists if `zil_free_lwb` races with either
`zil_commit_waiter_timeout` and/or `zil_lwb_flush_vdevs_done`.
Here's an example panic due to this bug:
> ::status
debugging crash dump vmcore.0 (64-bit) from ip-10-110-205-40
operating system: 5.11 dlpx-5.2.2.0_2017-12-04-17-28-32b6ba51fb (i86pc)
image uuid: 4af0edfb-e58e-6ed8-cafc-d3e9167c7513
panic message:
BAD TRAP: type=e (#pf Page fault) rp=ffffff0010555970 addr=60 occurred in module "zfs" due to a NULL pointer dereference
dump content: kernel pages only
> $c
zio_shrink+0x12()
zil_lwb_write_issue+0x30d(ffffff03dcd15cc0, ffffff03e0730e20)
zil_commit_waiter_timeout+0xa2(ffffff03dcd15cc0, ffffff03d97ffcf8)
zil_commit_waiter+0xf3(ffffff03dcd15cc0, ffffff03d97ffcf8)
zil_commit+0x80(ffffff03dcd15cc0, 9a9)
zfs_write+0xc34(ffffff03dc38b140, ffffff0010555e60, 40, ffffff03e00fb758, 0)
fop_write+0x5b(ffffff03dc38b140, ffffff0010555e60, 40, ffffff03e00fb758, 0)
write+0x250(42, fffffd7ff4832000, 2000)
sys_syscall+0x177()
If there's an outstanding lwb that's in `zil_commit_waiter_timeout`
waiting to timeout, waiting on it's waiter's CV, we must be sure not to
call `zil_free_lwb`. If we end up calling `zil_free_lwb`, then that LWB
may be freed and can result in a use-after-free situation where the
stale lwb pointer stored in the `zil_commit_waiter_t` structure of the
thread waiting on the waiter's CV is used.
A similar situation can occur if an lwb is issued to disk, and thus in
the `LWB_STATE_ISSUED` state, and `zil_free_lwb` is called while the
disk is servicing that lwb. In this situation, the lwb will be freed by
`zil_free_lwb`, which will result in a use-after-free situation when the
lwb's zio completes, and `zil_lwb_flush_vdevs_done` is called.
This race condition is prevented in `zil_close` by calling `zil_commit`
before `zil_free_lwb` is called, which will ensure all outstanding (i.e.
all lwb's in the `LWB_STATE_OPEN` and/or `LWB_STATE_ISSUED` states)
reach the `LWB_STATE_DONE` state before the lwb's are freed
(`zil_commit` will not return untill all the lwb's are
`LWB_STATE_DONE`).
Further, this race condition is prevented in `zil_sync` by only calling
`zil_free_lwb` for lwb's that do not have their `lwb_buf` pointer set.
All lwb's not in the `LWB_STATE_DONE` state will have a non-null value
for this pointer; the pointer is only cleared in
`zil_lwb_flush_vdevs_done`, at which point the lwb's state will be
changed to `LWB_STATE_DONE`.
This race is present in `zil_suspend`, leading to this bug.
At first glance, it would appear as though this would not be true
because `zil_suspend` will call `zil_commit`, just like `zil_close`, but
the problem is that `zil_suspend` will set the zilog's `zl_suspend`
field prior to calling `zil_commit`. Further, in `zil_commit`, if
`zl_suspend` is set, `zil_commit` will take a special branch of logic
and use `txg_wait_synced` instead of performing the normal `zil_commit`
logic.
This call to `txg_wait_synced` might be good enough for the data to
reach disk safely before it returns, but it does not ensure that all
outstanding lwb's reach the `LWB_STATE_DONE` state before it returns.
This is because, if there's an lwb "stuck" in
`zil_commit_waiter_timeout`, waiting for it's lwb to timeout, it will
maintain a non-null value for it's `lwb_buf` field and thus `zil_sync`
will not free that lwb. Thus, even though the lwb's data is already on
disk, the lwb will be left lingering, waiting on the CV, and will
eventually timeout and be issued to disk even though the write is
unnesseary.
So, after `zil_commit` is called from `zil_suspend`, we incorrectly
assume that there are not outstanding lwb's, and proceed to free all
lwb's found on the zilog's lwb list. As a result, we free the lwb that
will later be used `zil_commit_waiter_timeout`.
Reviewed by: John Kennedy <jwk404@gmail.com>
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Reviewed by: George Wilson <george.wilson@delphix.com>
Reviewed by: Brad Lewis <brad.lewis@delphix.com>
Reviewed by: Igor Kozhukhov <igor@dilos.org>
Approved by: Robert Mustacchi <rm@joyent.com>
Author: Prakash Surya <prakash.surya@delphix.com>
illumos/illumos-gate@a3b2868063https://www.illumos.org/issues/8677
We want to be able to run channel programs outside of synching context.
This would greatly improve performance of channel program that just gather
information, as we won't have to wait for synching context anymore.
This feature should introduce the following:
- A new command line flag in "zfs program" to specify our intention to
run in open context.
- A new flag/option within the channel program ioctl which selects the
context.
- Appropriate error handling whenever we try a channel program in
open-context that contains zfs.sync* expressions.
- Documentation for the new feature in the manual pages.
Reviewed by: Matt Ahrens <mahrens@delphix.com>
Reviewed by: Chris Williamson <chris.williamson@delphix.com>
Reviewed by: Pavel Zakharov <pavel.zakharov@delphix.com>
Approved by: Robert Mustacchi <rm@joyent.com>
Author: Serapheim Dimitropoulos <serapheim@delphix.com>
