Signed-off-by: Anton Gubarkov <anton.gubarkov@gmail.com>
Reviewed-by: Tony Hutter <hutter2@llnl.gov>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
`zpool_do_import()` passes `argv[0]`, (optionally) `argv[1]`, and
`pool_specified` to `import_pools()`. If `pool_specified==FALSE`, the
`argv[]` arguments are not used. However, these values may be off the
end of the `argv[]` array, so loading them could dereference unmapped
memory. This error is reported by the asan build:
```
=================================================================
==6003==ERROR: AddressSanitizer: heap-buffer-overflow
READ of size 8 at 0x6030000004a8 thread T0
#0 0x562a078b50eb in zpool_do_import zpool_main.c:3796
#1 0x562a078858c5 in main zpool_main.c:10709
#2 0x7f5115231bf6 in __libc_start_main
#3 0x562a07885eb9 in _start
0x6030000004a8 is located 0 bytes to the right of 24-byte region
allocated by thread T0 here:
#0 0x7f5116ac6b40 in __interceptor_malloc
#1 0x562a07885770 in main zpool_main.c:10699
#2 0x7f5115231bf6 in __libc_start_main
```
This commit passes NULL for these arguments if they are off the end
of the `argv[]` array.
Reviewed-by: George Wilson <gwilson@delphix.com>
Reviewed-by: John Kennedy <john.kennedy@delphix.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Allan Jude <allan@klarasystems.com>
Signed-off-by: Matthew Ahrens <mahrens@delphix.com>
Closes#12339
Use strlcpy instead of problematic strncpy
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Richard Elling <Richard.Elling@RichardElling.com>
Reviewed-by: Ryan Moeller <ryan@iXsystems.com>
Signed-off-by: George Melikov <mail@gmelikov.ru>
Closes#12344
zfs-send(8) claimed in the flags list you could use -pR when sending
a readonly filesystem or volume. You cannot.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Tony Nguyen <tony.nguyen@delphix.com>
Signed-off-by: Rich Ercolani <rincebrain@gmail.com>
Closes#12336
Currently, there are several places in zvol_id where the program logic
returns particular errno values, or even particular ioctl return values,
as the program exit status, rather than a straightforward system of
explicit zero on success and explicit nonzero value(s) on failure.
This is problematic for multiple reasons. One particularly interesting
problem that can arise, is that if any of these values happens to have
all 8 least significant bits unset (i.e., it is a positive or negative
multiple of 256), then although the C program sees a nonzero int value
(presumed to be a failure exit status), the actual exit status as seen
by the system is only the bottom 8 bits of that integer: zero.
This can happen in practice, and I have encountered it myself. In a
particularly weird situation, the zvol_open code in the zfs kernel
module was behaving in such a manner that it caused the open() syscall
to fail and for errno to be set to a kernel-private value (ERESTARTSYS,
which happens to be defined as 512). It turns out that 512 is evenly
divisible by 256; or, in other words, its least significant 8 bits are
all-zero. So even though zvol_id believed it was returning a nonzero
(failure) exit status of 512, the system modulo'd that value by 256,
resulting in the actual exit status visible by other programs being 0!
This actually-zero (non-failure) exit status caused problems: udev
believed that the program was operating successfully, when in fact it
was attempting to indicate failure via a nonzero exit status integer.
Combined with another problem, this led to the creation of nonsense
symlinks for zvol dev nodes by udev.
Let's get rid of all this problematic logic, and simply return
EXIT_SUCCESS (0) is everything went fine, and EXIT_FAILURE (1) if
anything went wrong.
Additionally, let's clarify some of the variable names (error is similar
to errno, etc) and clean up the overall program flow a bit.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Pavel Zakharov <pavel.zakharov@delphix.com>
Signed-off-by: Justin Gottula <justin@jgottula.com>
Closes#12302
The zvol_id program is invoked by udev, via a PROGRAM key in the
60-zvol.rules.in rule file, to determine the "pretty" /dev/zvol/*
symlink paths paths that should be generated for each opaquely named
/dev/zd* dev node.
The udev rule uses the PROGRAM key, followed by a SYMLINK+= assignment
containing the %c substitution, to collect the program's stdout and then
"paste" it directly into the name of the symlink(s) to be created.
Unfortunately, as currently written, zvol_id outputs both its intended
output (a single string representing the symlink path that should be
created to refer to the name of the dataset whose /dev/zd* path is
given) AND its error messages (if any) to stdout.
When processing PROGRAM keys (and others, such as IMPORT{program}), udev
uses only the data written to stdout for functional purposes. Any data
written to stderr is used solely for the purposes of logging (if udev's
log_level is set to debug).
The unintended consequence of this is as follows: if zvol_id encounters
an error condition; and then udev fails to halt processing of the
current rule (either because zvol_id didn't return a nonzero exit
status, or because the PROGRAM key in the rule wasn't written properly
to result in a "non-match" condition that would stop the current rule on
a nonzero exit); then udev will create a space-delimited list of symlink
names derived directly from the words of the error message string!
I've observed this exact behavior on my own system, in a situation where
the open() syscall on /dev/zd* dev nodes was failing sporadically (for
reasons that aren't especially relevant here). Because the open() call
failed, zvol_id printed "Unable to open device file: /dev/zd736\n" to
stdout and then exited.
The udev rule finished with SYMLINK+="zvol/%c %c". Assuming a volume
name like pool/foo/bar, this would ordinarily expand to
SYMLINK+="zvol/pool/foo/bar pool/foo/bar"
and would cause symlinks to be created like this:
/dev/zvol/pool/foo/bar -> /dev/zd736
/dev/pool/foo/bar -> /dev/zd736
But because of the combination of error messages being printed to
stdout, and the udev syntax freely accepting a space-delimited sequence
of names in this context, the error message string
"Unable to open device file: /dev/zd736\n"
in reality expanded to
SYMLINK+="zvol/Unable to open device file: /dev/zd736"
which caused the following symlinks to actually be created:
/dev/zvol/Unable -> /dev/zd736
/dev/to -> /dev/zd736
/dev/open -> /dev/zd736
/dev/device -> /dev/zd736
/dev/file: -> /dev/zd736
/dev//dev/zd736 -> /dev/zd736
(And, because multiple zvols had open() syscall errors, multiple zvols
attempted to claim several of those symlink names, resulting in numerous
udev errors and timeouts and general chaos.)
This commit rectifies all this silliness by simply printing error
messages to stderr, as Dennis Ritchie originally intended.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Pavel Zakharov <pavel.zakharov@delphix.com>
Signed-off-by: Justin Gottula <justin@jgottula.com>
Closes#12302
This enables ZED to auto-online vdevs that are not wholedisk managed by
ZFS.
Signed-off-by: Ryan Moeller <ryan@iXsystems.com>
Reviewed-by: Don Brady <don.brady@delphix.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Tony Hutter <hutter2@llnl.gov>
Commit 6fc3099 broke the quoting when invoking the mail program, revert
that change.
Signed-off-by: Laurențiu Nicola <lnicola@dend.ro>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Tony Hutter <hutter2@llnl.gov>
ZFS loves using %llu for uint64_t, but that requires a cast to not
be noisy - which is even done in many, though not all, places.
Also a couple places used %u for uint64_t, which were promoted
to %llu.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Rich Ercolani <rincebrain@gmail.com>
Closes#12233
This replaces the generic libspl atomic.c atomics implementation
with one based on builtin gcc atomics. This functionality was added
as an experimental feature in gcc 4.4. Today even CentOS 7 ships
with gcc 4.8 as the default compiler we can make this the default.
Furthermore, the builtin atomics are as good or better than our
hand-rolled implementation so it's reasonable to drop that custom code.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Ahelenia Ziemiańska <nabijaczleweli@nabijaczleweli.xyz>
Closes#11904Closes#12252Closes#12244
Starting in Linux 5.10, trying to write to /dev/{null,zero} errors out.
Prefer to inform people when this happens rather than hoping they guess
what's wrong.
Reviewed-by: Antonio Russo <aerusso@aerusso.net>
Reviewed-by: Ahelenia Ziemiańska <nabijaczleweli@nabijaczleweli.xyz>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: John Kennedy <john.kennedy@delphix.com>
Signed-off-by: Rich Ercolani <rincebrain@gmail.com>
Closes: #11991
The first warning of a misspelling is a false positive, so we annotate
the script accordingly. As for the x-prefix warnings update the check
to use the conventional '[ -z <string> ]' syntax.
all-syslog.sh:46:47: warning: Possible misspelling: ZEVENT_ZIO_OBJECT
may not be assigned, but ZEVENT_ZIO_OBJSET is. [SC2153]
make_gitrev.sh:53:6: note: Avoid x-prefix in comparisons as it no
longer serves a purpose [SC2268]
man-dates.sh:10:7: note: Avoid x-prefix in comparisons as it no
longer serves a purpose [SC2268]
Reviewed-by: Ahelenia Ziemiańska <nabijaczleweli@nabijaczleweli.xyz>
Reviewed-by: John Kennedy <john.kennedy@delphix.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#12208
This:
(a) improves the error log message,
(b) locks per pool instead of globally,
(c) locks the actual output file instead of /var/lock/zfs-list,
which would otherwise linger there forever (well, still will,
but you can remove it and it won't come back), and
(d) preserves attributes of the output file
instead of reverting them to 0:0 644
It is imperative that the previous commit
("zed-functions.sh: zed_lock(): don't truncate lock")
be included in any series that contains this one
Reviewed-by: Tony Hutter <hutter2@llnl.gov>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Ahelenia Ziemiańska <nabijaczleweli@nabijaczleweli.xyz>
Closes#12042
By locking the log file itself, we can omit arduous rebinding and
explicit umask setting, but, perhaps more importantly, avoid permanently
littering /var/lock/ with zed.debug.log.lock we will never delete
It is imperative that the previous commit
("zed-functions.sh: zed_lock(): don't truncate lock")
be included in any series that contains this one
Reviewed-by: Tony Hutter <hutter2@llnl.gov>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Ahelenia Ziemiańska <nabijaczleweli@nabijaczleweli.xyz>
Closes#12042
By appending instead of truncating, we can lock on any file (with write
permissions) instead of only dedicated lock files, since the locking
process itself no longer alters the file in any way
Reviewed-by: Tony Hutter <hutter2@llnl.gov>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Ahelenia Ziemiańska <nabijaczleweli@nabijaczleweli.xyz>
Closes#12042
Update the logic to handle the dedup-case of consecutive
FREEs in the livelist code. The logic still ensures that
all the FREE entries are matched up with a respective
ALLOC by keeping a refcount for each FREE blkptr that we
encounter and ensuring that this refcount gets to zero
by the time we are done processing the livelist.
zdb -y no longer panics when encountering double frees
Reviewed-by: Matthew Ahrens <mahrens@delphix.com>
Reviewed-by: John Kennedy <john.kennedy@delphix.com>
Reviewed-by: Don Brady <don.brady@delphix.com>
Signed-off-by: Serapheim Dimitropoulos <serapheim@delphix.com>
Closes#11480Closes#12177
It turns out that sometimes, evidently only when run inside the
ZTS handler, arc_summary3 | head > /dev/null will die with ENOTCONN,
and ruin the test run.
Added handling for that.
Reviewed-by: John Kennedy <john.kennedy@delphix.com>
Reviewed-by: Ryan Moeller <ryan@iXsystems.com>
Signed-off-by: Rich Ercolani <rincebrain@gmail.com>
Closes#12160
There used to be a warning after upgrading a zpool in FreeBSD, so users
won't forget to update the boot loader that pool is booted from.
This change brings this warning back, but only if the bootfs property
is set on the pool, which should be sufficient for the vast majority of
FreeBSD installations. People running something custom are most likely
aware of what to do after an upgrade in their specific environment.
Functionality is implemented in an OS specific helper function.
Reviewed-by: John Kennedy <john.kennedy@delphix.com>
Reviewed-by: Ryan Moeller <ryan@iXsystems.com>
Co-authored-by: Michael Gmelin <grembo@FreeBSD.org>
Signed-off-by: Michael Gmelin <grembo@FreeBSD.org>
Closes#12099Closes#12104
make_gitrev.sh actually breaks checkbashisms' parser,
which /insists/ that the end-of-line " is actually a string start
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Ahelenia Ziemiańska <nabijaczleweli@nabijaczleweli.xyz>
Closes#12101
This checks every file it checked (and a few more),
but explicitly instead of "if it works it works" best-effort
(which wasn't that good anyway)
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Ahelenia Ziemiańska <nabijaczleweli@nabijaczleweli.xyz>
Closes#10512Closes#12101
Accidentally introduced by commit dd00925e8d.
Force-install the zstreamdump link, this is a supported configuration
and the install should not fail if it needs to overwrite an existing
file.
Also cd to work around some funny platforms as noted in AC_PROG_LN_S doc
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Ryan Moeller <ryan@iXsystems.com>
Reviewed-by: Allan Jude <allan@klarasystems.com>
Signed-off-by: Ahelenia Ziemiańska <nabijaczleweli@nabijaczleweli.xyz>
Closes#12143
This change introduces long options for ztest. It builds the usage
message as well as the long_options array from a single table. It also
adds #defines for the default values.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: John Kennedy <john.kennedy@delphix.com>
Signed-off-by: Manoj Joseph <manoj.joseph@delphix.com>
Closes#12117
zstreamdump(8) was in quite a bad state,
and the wrapper didn't work if invoked without /sbin in $PATH
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Ahelenia Ziemiańska <nabijaczleweli@nabijaczleweli.xyz>
Closes#12015
The change correctly handles BrokenPipeError and improves the
associated tests.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: John Kennedy <john.kennedy@delphix.com>
Signed-off-by: Rich Ercolani <rincebrain@gmail.com>
Closes#12037Closes#12036
Propagate vdev child state to parents on invalid label
Add VDEV_AUX_BAD_LABEL to print_import_config()
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Mark Maybee <mark.maybee@delphix.com>
Co-authored-by: Srikanth N S <srikanth.nagasubbaraoseetharaman@hpe.com>
Signed-off-by: Vipin Kumar Verma <vipin.verma@hpe.com>
Closes#12088
Afterward, git grep ZoL matches:
* README.md: * [ZoL Site](https://zfsonlinux.org)
- Correct
* etc/default/zfs.in:# ZoL userland configuration.
- Changing this would induce a needless upgrade-check,
if the user has modified the configuration;
this can be updated the next time the defaults change
* module/zfs/dmu_send.c: * ZoL < 0.7 does not handle [...]
- Before 0.7 is ZoL, so fair enough
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Ahelenia Ziemiańska <nabijaczleweli@nabijaczleweli.xyz>
Issue #11956
This can be very easily triggered by adding a sleep(1) before
the wait4() on a PID-starved system: the reaper thread would wait
for a child before its entry appeared, letting old entries accumulate:
Invoking "all-debug.sh" eid=3021 pid=391
Finished "(null)" eid=0 pid=391 time=0.002432s exit=0
Invoking "all-syslog.sh" eid=3021 pid=336
Finished "(null)" eid=0 pid=336 time=0.002432s exit=0
Invoking "history_event-zfs-list-cacher.sh" eid=3021 pid=347
Invoking "all-debug.sh" eid=3022 pid=349
Finished "history_event-zfs-list-cacher.sh" eid=3021 pid=347
time=0.001669s exit=0
Finished "(null)" eid=0 pid=349 time=0.002404s exit=0
Invoking "all-syslog.sh" eid=3022 pid=370
Finished "(null)" eid=0 pid=370 time=0.002427s exit=0
Invoking "history_event-zfs-list-cacher.sh" eid=3022 pid=391
avl_find(tree, new_node, &where) == NULL
ASSERT at ../../module/avl/avl.c:641:avl_add()
Thread 1 "zed" received signal SIGABRT, Aborted.
By employing this wider lock, we atomise [wait, remove] and [fork, add]:
slowing down the reaper thread now just causes some zombies
to accumulate until it can get to them
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Don Brady <don.brady@delphix.com>
Signed-off-by: Ahelenia Ziemiańska <nabijaczleweli@nabijaczleweli.xyz>
Closes#11963Closes#11965
Also minor clean-up with folding state_to_val() into a case,
unrolling the lesser-available seq into numbers,
ignoring vdev states we don't care about,
and documentation comments
Signed-off-by: Tony Hutter <hutter2@llnl.gov>
Signed-off-by: Ahelenia Ziemiańska <nabijaczleweli@nabijaczleweli.xyz>
Closes#11934Closes#11935
Under function map_slot() variable passed as args
were not getting properly substituted or expanded.
This patch fixes the substitution issue.
Reviewed-by: Niklas Edmundsson <nikke@acc.umu.se>
Reviewed-by: Tony Hutter <hutter2@llnl.gov>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Arshad Hussain <arshad.hussain@aeoncomputing.com>
Closes#11951Closes#11959
If zdb is not built with DEBUG mode, the ASSERT macros will be
eliminated.
This will leave vim defined, but not used (gcc warning) and
checkpoint spacemap validation loop will do nothing.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Toomas Soome <tsoome@me.com>
Closes#11932
As soon as wait4() returns, fork() can immediately return with the same
PID, and race to lock _launched_processes_lock, then try to add the new
(duplicate) PID to _launched_processes, which asserts
By locking before wait4(), we ensure, that, given that same
unfortunate scheduling, _launched_processes_lock cannot be locked by the
spawner before we pop the process in the reaper, and only afterward will
it be added
This moves where the reaper idles when there are children from the
wait4() to the pause(), locking for the duration of that single syscall
in both the no-children and running-children cases; the impact of this
is one to two syscalls (depending on _launched_processes_lock state)
per loop
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Don Brady <don.brady@delphix.com>
Signed-off-by: Ahelenia Ziemiańska <nabijaczleweli@nabijaczleweli.xyz>
Closes#11924Closes#11928
Also don't dup /dev/null over stdio if daemonised
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Ahelenia Ziemiańska <nabijaczleweli@nabijaczleweli.xyz>
Closes#11891
Do not (incorrectly, right instead left) pad health string itself,
it will be taken care of when printing property value below.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Ryan Moeller <ryan@iXsystems.com>
Signed-off-by: Yuri Pankov <yuripv@FreeBSD.org>
Closes#11899
Several improvements to the operation of the 'compatibility' property:
1) Improved handling of unrecognized features:
Change the way unrecognized features in compatibility files are handled.
* invalid features in files under /usr/share/zfs/compatibility.d
only get a warning (as these may refer to future features not yet in
the library),
* invalid features in files under /etc/zfs/compatibility.d
get an error (as these are presumed to refer to the current system).
2) Improved error reporting from zpool_load_compat.
Note: slight ABI change to zpool_load_compat for better error reporting.
3) compatibility=legacy inhibits all 'zpool upgrade' operations.
4) Detect when features are enabled outside current compatibility set
* zpool set compatibility=foo <-- print a warning
* zpool set feature@xxx=enabled <-- error
* zpool status <-- indicate this state
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Colm Buckley <colm@tuatha.org>
Closes#11861
A tentative implementation and discussion was done in #5285.
According to it a send --skip-missing|-s flag has been added.
In a replication stream, when there are snapshots missing in
the hierarchy, if -s is provided print a warning and ignore
dataset (and its children) instead of throwing an error
Reviewed-by: Paul Dagnelie <pcd@delphix.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Pablo Correa Gómez <ablocorrea@hotmail.com>
Closes#11710
list_zvols() would happily, for zvols with spaces in their names,
assign the second half to volmode, &c., so use a normal read
and set IFS to a tab instead of using 4 separate AWK processes(?)
Similarly, in filter_out_deleted_zvols(), run zfs(8) once and use the
output directly instead of spawning a zfs(8) process per zvol
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Pavel Zakharov <pavel.zakharov@delphix.com>
Reviewed-by: Ryan Moeller <ryan@iXsystems.com>
Signed-off-by: Ahelenia Ziemiańska <nabijaczleweli@nabijaczleweli.xyz>
Closes#11859
Dunno, maybe it's just me, but the previous style was /really/ confusing
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Ahelenia Ziemiańska <nabijaczleweli@nabijaczleweli.xyz>
Closes#11860
It's all of 40 bytes with 4-byte pointers and 64 with 8-byte ones
(previously 44 and 88, respectively) ‒
there's no reason it can't live on the stack
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Ahelenia Ziemiańska <nabijaczleweli@nabijaczleweli.xyz>
Closes#11860
No users, fields marked "reserved for future use", macros defined to 0
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Ahelenia Ziemiańska <nabijaczleweli@nabijaczleweli.xyz>
Closes#11860
No users, nobody sets it, main() hard-codes LOG_DAEMON, which is the
only correct value for this
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Ahelenia Ziemiańska <nabijaczleweli@nabijaczleweli.xyz>
Closes#11860
Users passed in EXIT_SUCCESS and EXIT_FAILURE, despite it being a bool
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Ahelenia Ziemiańska <nabijaczleweli@nabijaczleweli.xyz>
Closes#11860
We set SA_RESTART early on, which will prevent EINTRs (indeed, to the
point of needing to clear it in the reaper, since it interferes with
pause(2)), which is the only error zed_file_write_n() actually handled
(plus, the pid write is no bigger than 12 bytes anyway)
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Ahelenia Ziemiańska <nabijaczleweli@nabijaczleweli.xyz>
Closes#11834
These events should currently never be generated.
Also untag _zed_event_add_nvpair() from merge with
zpool_do_events_nvprint() ‒ they serve different purposes (machine,
usually script vs human consumption) and format the output differently
as it stands
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Ahelenia Ziemiańska <nabijaczleweli@nabijaczleweli.xyz>
Closes#11834
Same deal as zed_file_close_on_exec()
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Ahelenia Ziemiańska <nabijaczleweli@nabijaczleweli.xyz>
Closes#11834
And add a note on /why/ ZEDLETs need to be owned by root
Quoth chown(2), Linux man-pages project:
Only a privileged process (Linux: one with the CAP_CHOWN capability)
may change the owner of a file.
Quoth chown(2), FreeBSD:
[EPERM] The operation would change the ownership,
but the effective user ID is not the super-user.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Ahelenia Ziemiańska <nabijaczleweli@nabijaczleweli.xyz>
Closes#11834
There simply isn't a need for one, since the flags the daemon takes
are all short (mostly just toggles) and administrative in nature,
and are therefore better served by the age-old tradition of sourcing an
environment file and preparing the cmdline in the init-specific handler
itself, if needed at all
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Ahelenia Ziemiańska <nabijaczleweli@nabijaczleweli.xyz>
Closes#11834
/dev/fd on Darwin
Consider the following strace output:
prlimit64(0, RLIMIT_NOFILE, NULL, {rlim_cur=1024, rlim_max=1024*1024}) = 0
Yes, that is well over a million file descriptors!
This reduces the ZED start-up time from "at least a second" to
"instantaneous", and, under strace, from "don't even try" to "usable"
by simple virtue of doing five syscalls instead of over a million;
in most cases the main loop does nothing
Recent Linuxes (5.8+) have close_range(2) for this, but that's an
overoptimisation (and libcs don't have wrappers for it yet)
This is also run by the ZEDLET pre-exec. Compare:
Finished "all-syslog.sh" eid=13 pid=6717 time=1.027100s exit=0
Finished "history_event-zfs-list-cacher.sh" eid=13 pid=6718 time=1.046923s exit=0
to
Finished "all-syslog.sh" eid=12 pid=4834 time=0.001836s exit=0
Finished "history_event-zfs-list-cacher.sh" eid=12 pid=4835 time=0.001346s exit=0
lol
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Ahelenia Ziemiańska <nabijaczleweli@nabijaczleweli.xyz>
Closes#11834
200ms time-out is relatively long, but if we already hit the cap,
then we'll likely be able to spawn multiple new jobs when we wake up
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Ahelenia Ziemiańska <nabijaczleweli@nabijaczleweli.xyz>
Closes#11807
The FIXME comment was there since the initial implementation in 2014,
there are no users
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Ahelenia Ziemiańska <nabijaczleweli@nabijaczleweli.xyz>
Closes#11807
zpool list, which is the only user, would mistakenly try to parse the
empty string as the interval in this case:
$ zpool list "a"
cannot open 'a': no such pool
$ zpool list ""
interval cannot be zero
usage: <usage string follows>
which is now symmetric with zpool get:
$ zpool list ""
cannot open '': name must begin with a letter
Avoid breaking the "interval cannot be zero" string.
There simply isn't a need for this, and it's user-facing.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Ryan Moeller <ryan@iXsystems.com>
Signed-off-by: Ahelenia Ziemiańska <nabijaczleweli@nabijaczleweli.xyz>
Closes#11841Closes#11843
Correct an assortment of typos throughout the code base.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Matthew Ahrens <mahrens@delphix.com>
Reviewed-by: Ryan Moeller <ryan@iXsystems.com>
Signed-off-by: Andrea Gelmini <andrea.gelmini@gelma.net>
Closes#11774
Update the fsck.zfs helper to bubble up some already-known-about
errors if they are detected in the pool.
health=degraded => 4/"Filesystem errors left uncorrected"
health=faulted && dataset in /etc/fstab => 8/"Operational error"
pool not found => 8/"Operational error"
everything else => 0
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Ahelenia Ziemiańska <nabijaczleweli@nabijaczleweli.xyz>
Closes#11806
ZoL 0.6.1 introduced feature flags with the three features that all
implementations at the time were guaranteed to have. 0.6.4 introduced
a few more until 0.6.5 added two after that. OpenZFS 2.1 added the
dRAID feature.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Mike Swanson <mikeonthecomputer@gmail.com>
Closes#11818
When a child process is killed waitpid() must be called on the
pid the reap the zombie process.
Update BUGS section to reflect reality by replacing "zedlets
aren't time limited with "zedlets can be interrupted".
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Ahelenia Ziemiańska <nabijaczleweli@nabijaczleweli.xyz>
Closes#11769Closes#11798
Reviewed-by: Matthew Ahrens <mahrens@delphix.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Andrea Gelmini <andrea.gelmini@gelma.net>
Closes#11775
If TX_WRITE is create on a file, and the file is later deleted and a new
directory is created on the same object id, it is possible that when
zil_commit happens, zfs_get_data will be called on the new directory.
This may result in panic as it tries to do range lock.
This patch fixes this issue by record the generation number during
zfs_log_write, so zfs_get_data can check if the object is valid.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Chunwei Chen <david.chen@nutanix.com>
Closes#10593Closes#11682
The RAIDZ and DRAID code is responsible for reporting checksum errors on
their child vdevs. Checksum errors represent events where a disk
returned data or parity that should have been correct, but was not. In
other words, these are instances of silent data corruption. The
checksum errors show up in the vdev stats (and thus `zpool status`'s
CKSUM column), and in the event log (`zpool events`).
Note, this is in contrast with the more common "noisy" errors where a
disk goes offline, in which case ZFS knows that the disk is bad and
doesn't try to read it, or the device returns an error on the requested
read or write operation.
RAIDZ/DRAID generate checksum errors via three code paths:
1. When RAIDZ/DRAID reconstructs a damaged block, checksum errors are
reported on any children whose data was not used during the
reconstruction. This is handled in `raidz_reconstruct()`. This is the
most common type of RAIDZ/DRAID checksum error.
2. When RAIDZ/DRAID is not able to reconstruct a damaged block, that
means that the data has been lost. The zio fails and an error is
returned to the consumer (e.g. the read(2) system call). This would
happen if, for example, three different disks in a RAIDZ2 group are
silently damaged. Since the damage is silent, it isn't possible to know
which three disks are damaged, so a checksum error is reported against
every child that returned data or parity for this read. (For DRAID,
typically only one "group" of children is involved in each io.) This
case is handled in `vdev_raidz_cksum_finish()`. This is the next most
common type of RAIDZ/DRAID checksum error.
3. If RAIDZ/DRAID is not able to reconstruct a damaged block (like in
case 2), but there happens to be additional copies of this block due to
"ditto blocks" (i.e. multiple DVA's in this blkptr_t), and one of those
copies is good, then RAIDZ/DRAID compares each sector of the data or
parity that it retrieved with the good data from the other DVA, and if
they differ then it reports a checksum error on this child. This
differs from case 2 in that the checksum error is reported on only the
subset of children that actually have bad data or parity. This case
happens very rarely, since normally only metadata has ditto blocks. If
the silent damage is extensive, there will be many instances of case 2,
and the pool will likely be unrecoverable.
The code for handling case 3 is considerably more complicated than the
other cases, for two reasons:
1. It needs to run after the main raidz read logic has completed. The
data RAIDZ read needs to be preserved until after the alternate DVA has
been read, which necessitates refcounts and callbacks managed by the
non-raidz-specific zio layer.
2. It's nontrivial to map the sections of data read by RAIDZ to the
correct data. For example, the correct data does not include the parity
information, so the parity must be recalculated based on the correct
data, and then compared to the parity that was read from the RAIDZ
children.
Due to the complexity of case 3, the rareness of hitting it, and the
minimal benefit it provides above case 2, this commit removes the code
for case 3. These types of errors will now be handled the same as case
2, i.e. the checksum error will be reported against all children that
returned data or parity.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Matthew Ahrens <mahrens@delphix.com>
Closes#11735
Importing a pool using the cachefile is ideal to reduce the time
required to import a pool. However, if the devices associated with
a pool in the cachefile have changed, then the import would fail.
This can easily be corrected by doing a normal import which would
then read the pool configuration from the labels.
The goal of this change is make importing using a cachefile more
resilient and auto-correcting. This is accomplished by having
the cachefile import logic automatically fallback to reading the
labels of the devices similar to a normal import. The main difference
between the fallback logic and a normal import is that the cachefile
import logic will only look at the device directories that were
originally used when the cachefile was populated. Additionally,
the fallback logic will always import by guid to ensure that only
the pools in the cachefile would be imported.
External-issue: DLPX-71980
Reviewed-by: Matthew Ahrens <mahrens@delphix.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: George Wilson <gwilson@delphix.com>
Closes#11716
vdev_id uses the /dev/mapper/ symlinks to resolve a UUID to a dm name
(like dm-1). However on some multipath setups, there is no /dev/mapper/
entry for the UUID at the time vdev_id is called by udev. However,
this isn't necessarily needed, as we may be able to resolve the dm
name from the $DEVNAME that udev passes us (like DEVNAME="/dev/dm-1").
This patch tries to resolve the dm name from $DEVNAME first, before
falling back to looking in /dev/mapper/. This fixed an issue where the
by-vdev names weren't reliably showing up on one of our nodes.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Tony Hutter <hutter2@llnl.gov>
Closes#11698
For some reason cppcheck 1.90 is generating an invalidSyntax warning
when the BF64_SET macro is used in the zstream source. The same
warning is not reported by cppcheck 2.3, nor is their any evident
problem with the expanded macro. This appears to be an issue with
this version of cppcheck. This commit annotates the source to suppress
the warning.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#11700
Bring the output of the removal status in line with the other
"fields" that zpool status outputs, and thus allows an parser to
easier detect this as continuation of the 'remove:' output.
Before:
remove: Removal of vdev 0 copied 282G in 0h9m, completed on [...]
776K memory used for removed device mappings
Now:
remove: Removal of vdev 0 copied 282G in 0h9m, completed on [...]
776K memory used for removed device mappings
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
Closes#11674
After 35ec517 it has become possible to import ZFS pools witn an
active org.illumos:edonr feature on FreeBSD, leading to a panic.
In addition, "zpool status" reported all pools without edonr
as upgradable and "zpool upgrade -v" reported edonr in the list
of upgradable features.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Ryan Moeller <ryan@iXsystems.com>
Signed-off-by: Martin Matuska <mm@FreeBSD.org>
Closes#11653
Given a DM device name, the old vdev_id script would extract any text
after a 'p' as the partition number. It then appends "-part" + the
partition number to the name, giving a by-vdev name like "L0-part5".
This works fine if the DM name is like 'dm-2p5', but doesn't work if
the DM name is a multipath name like "mpatha". In those cases it
incorrectly matches the 'p' in "mpatha", giving by-vdev names like
"L0-partatha".
This patch fixes the issue by making the partition regex match stricter.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Tony Hutter <hutter2@llnl.gov>
Closes#11637
gmake install fails when zpool.d compat links already exist.
Force the symlinks to be recreated if already present.
Reviewed-by: Matthew Ahrens <mahrens@delphix.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Ryan Moeller <ryan@iXsystems.com>
Closes#11633
I think this is the behavior that most users expect.
Future work: have a separate flag, e.g., -O, to specify separate
set_global_vars for the zdb child than for the ztest children.
Reviewed-by: Matthew Ahrens <mahrens@delphix.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Pavel Zakharov <pavel.zakharov@delphix.com>
Signed-off-by: Christian Schwarz <me@cschwarz.com>
Closes#11602
Without set_global_var() in the child processes the -o option provides
little use.
Before this change set_global_var() was called as a side-effect of
getopt processing which only happens for the parent ztest process.
This change limits the set of options that can be set and makes them
available to the child through ztest_shared_opts_t.
Future work: support arbitrary option count and length.
Reviewed-by: Matthew Ahrens <mahrens@delphix.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Pavel Zakharov <pavel.zakharov@delphix.com>
Signed-off-by: Christian Schwarz <me@cschwarz.com>
Closes#11602
That happens because of an off-by-one mistake.
share_mount_one_cb() calls report_mount_progress(current=sm_done) after
having incremented sm_done by one. Then report_mount_progress()
increments the parameter again. It appears that that logic became
obsolete after commit a10d50f999, parallel zfs mount.
On FreeBSD I observe that zfs mount -a -v prints, for example,
(null): (201/248)
That happens because set_progress_header() is never called.
With this change the output becomes correct:
Mounting ZFS filesystems: (209/248)
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Andriy Gapon <avg@FreeBSD.org>
Closes#11607
Property to allow sets of features to be specified; for compatibility
with specific versions / releases / external systems. Influences
the behavior of 'zpool upgrade' and 'zpool create'. Initial man
page changes and test cases included.
Brief synopsis:
zpool create -o compatibility=off|legacy|file[,file...] pool vdev...
compatibility = off : disable compatibility mode (enable all features)
compatibility = legacy : request that no features be enabled
compatibility = file[,file...] : read features from specified files.
Only features present in *all* files will be enabled on the
resulting pool. Filenames may be absolute, or relative to
/etc/zfs/compatibility.d or /usr/share/zfs/compatibility.d (/etc
checked first).
Only affects zpool create, zpool upgrade and zpool status.
ABI changes in libzfs:
* New function "zpool_load_compat" to load and parse compat sets.
* Add "zpool_compat_status_t" typedef for compatibility parse status.
* Add ZPOOL_PROP_COMPATIBILITY to the pool properties enum
* Add ZPOOL_STATUS_COMPATIBILITY_ERR to the pool status enum
An initial set of base compatibility sets are included in
cmd/zpool/compatibility.d, and the Makefile for cmd/zpool is
modified to install these in $pkgdatadir/compatibility.d and to
create symbolic links to a reasonable set of aliases.
Reviewed-by: ericloewe
Reviewed-by: Matthew Ahrens <mahrens@delphix.com>
Reviewed-by: Richard Laager <rlaager@wiktel.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Colm Buckley <colm@tuatha.org>
Closes#11468
Rather than conditionally compiling out the edonr code for FreeBSD
update zfs_mod_supported_feature() to indicate this feature is
unsupported. This ensures that all spa features are defined on
every platform, even if they are not supported.
Reviewed-by: Ryan Moeller <ryan@ixsystems.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#11605
Issue #11468
There are two issues that don't allow ZFS to be compiled using uClibc.
`backtrace()`, and `program_invocation_short_name` as a `const`.
This patch adds uClibc to the conditionals in the same way there are
already for Glibc for `backtrace()`; and removes the external param
`program_invocation_short_name` because its only used here for the
whole project.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: José Luis Salvador Rufo <salvador.joseluis@gmail.com>
Closes#11600
Within function sas_handler() userspace commands like
'/usr/sbin/multipath' have been replaced with sourcing
device details from within sysfs which reduced a
significant amount of overhead and processing time.
Multiple JBOD enclosures and their order are sourced
from the bsg driver (/sys/class/enclosure) to isolate
chassis top-level expanders, which are then dynamically
indexed based on host channel of the multipath subordinate
disk member device being processed. Additionally added a
"mixed" mode for slot identification for environments where
a ZFS server system may contain SAS disk slots where there
is no expander (direct connect to HBA) while an attached
external JBOD with an expander have different slot identifier
methods.
How Has This Been Tested?
~~~~~~~~~~~~~~~~~~~~~~~~~
Testing was performed on a AMD EPYC based dual-server
high-availability multipath environment with multiple
HBAs per ZFS server and four SAS JBODs. The two primary
JBODs were multipath/cross-connected between the two
ZFS-HA servers. The secondary JBODs were daisy-chained
off of the primary JBODs using aligned SAS expander
channels (JBOD-0 expanderA--->JBOD-1 expanderA,
JBOD-0 expanderB--->JBOD-1 expanderB, etc).
Pools were created, exported and re-imported, imported
globally with 'zpool import -a -d /dev/disk/by-vdev'.
Low level udev debug outputs were traced to isolate
and resolve errors.
Result:
~~~~~~~
Initial testing of a previous version of this change
showed how reliance on userspace utilities like
'/usr/sbin/multipath' and '/usr/bin/lsscsi' were
exacerbated by increasing numbers of disks and JBODs.
With four 60-disk SAS JBODs and 240 disks the time to
process a udevadm trigger was 3 minutes 30 seconds
during which nearly all CPU cores were above 80%
utilization. By switching reliance on userspace
utilities to sysfs in this version, the udevadm
trigger processing time was reduced to 12.2 seconds
and negligible CPU load.
This patch also fixes few shellcheck complains.
Reviewed-by: Gabriel A. Devenyi <gdevenyi@gmail.com>
Reviewed-by: Tony Hutter <hutter2@llnl.gov>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Co-authored-by: Jeff Johnson <jeff.johnson@aeoncomputing.com>
Signed-off-by: Jeff Johnson <jeff.johnson@aeoncomputing.com>
Signed-off-by: Arshad Hussain <arshad.hussain@aeoncomputing.com>
Closes#11526
The pool guid and vdev guid received by zfs_agent_post_event(),
which calls zfs_retire_recv(), are normally non-zero. However,
later in this same method they may be unconditionally reset to
zero by the code which is intended to handle multipath, spare
and l2arc vdevs. This will result in the EC_dev_remove not
being handled.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>\
Co-authored-by: Vipin Kumar Verma <vipin.verma@hpe.com>
Signed-off-by: Srikanth N S <srikanth.nagasubbaraoseetharaman@hpe.com>
Closes#11564
In ZED zfs_retire agent added a check to handle Distributed Spare
replacement for Faulted VDEV also.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Co-authored-by: Vipin Kumar Verma <vipin.verma@hpe.com>
Signed-off-by: Mark Maybee <mark.maybee@hpe.com>
Closes#11354Closes#11355
While you can use zdb -R poolname vdev:offset:[<lsize>/]<psize>[:flags]
to extract individual DVAs from a vdev, it would be handy for be able
copy an entire file out of the pool.
Given a file or object number, add support to copy the contents to a
file. Useful for debugging and recovery.
Reviewed-by: Jorgen Lundman <lundman@lundman.net>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Allan Jude <allan@klarasystems.com>
Closes#11027
In order for cppcheck to perform a proper analysis it needs to be
aware of how the sources are compiled (source files, include
paths/files, extra defines, etc). All the needed information is
available from the Makefiles and can be leveraged with a generic
cppcheck Makefile target. So let's add one.
Additional minor changes:
* Removing the cppcheck-suppressions.txt file. With cppcheck 2.3
and these changes it appears to no longer be needed. Some inline
suppressions were also removed since they appear not to be
needed. We can add them back if it turns out they're needed
for older versions of cppcheck.
* Added the ax_count_cpus m4 macro to detect at configure time how
many processors are available in order to run multiple cppcheck
jobs. This value is also now used as a replacement for nproc
when executing the kernel interface checks.
* "PHONY =" line moved in to the Rules.am file which is included
at the top of all Makefile.am's. This is just convenient becase
it allows us to use the += syntax to add phony targets.
* One upside of this integration worth mentioning is it now allows
`make cppcheck` to be run in any directory to check that subtree.
* For the moment, cppcheck is not run against the FreeBSD specific
kernel sources. The cppcheck-FreeBSD target will need to be
implemented and testing on FreeBSD to support this.
Reviewed-by: Ryan Moeller <ryan@ixsystems.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#11508
Explicitly check for NULL to satisfy cppcheck that "val" can never
be NULL when passed to printf(). This looks like a false positive
since is_blank_str() can never take the false conditional branch
when passed a NULL. But there's no harm in adding the extra check.
Reviewed-by: Ryan Moeller <ryan@ixsystems.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#11508
It was observed that vdev_id exists silently when
the $CONFIG file is missing.
This patch adds error message in case vdev_id is
called without default $CONFIG or '-c'. This makes
end user observe the exit message more easily.
Before Patch:
~~~~~~~~~~~~~
$ ./cmd/vdev_id/vdev_id
$
After Patch:
~~~~~~~~~~~~
$ ./cmd/vdev_id/vdev_id
Error: Config file "/etc/zfs/vdev_id.conf" not found
$
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Arshad Hussain <arshad.hussain@aeoncomputing.com>
Closes#11498
When creating a pool only features supported by both user and
kernel space should be enabled. Furthermore, improve the error
messages when attempting to create, or add, a dRAID vdev when
the dRAID feature is not supported by the kernel modules.
Reviewed-by: Mark Maybee <mark.maybee@delphix.com>
Reviewed-by: Matthew Ahrens <mahrens@delphix.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes#11492
Mixing ZIL and normal allocations has several problems:
1. The ZIL allocations are allocated, written to disk, and then a few
seconds later freed. This leaves behind holes (free segments) where the
ZIL blocks used to be, which increases fragmentation, which negatively
impacts performance.
2. When under moderate load, ZIL allocations are of 128KB. If the pool
is fairly fragmented, there may not be many free chunks of that size.
This causes ZFS to load more metaslabs to locate free segments of 128KB
or more. The loading happens synchronously (from zil_commit()), and can
take around a second even if the metaslab's spacemap is cached in the
ARC. All concurrent synchronous operations on this filesystem must wait
while the metaslab is loading. This can cause a significant performance
impact.
3. If the pool is very fragmented, there may be zero free chunks of
128KB or more. In this case, the ZIL falls back to txg_wait_synced(),
which has an enormous performance impact.
These problems can be eliminated by using a dedicated log device
("slog"), even one with the same performance characteristics as the
normal devices.
This change sets aside one metaslab from each top-level vdev that is
preferentially used for ZIL allocations (vdev_log_mg,
spa_embedded_log_class). From an allocation perspective, this is
similar to having a dedicated log device, and it eliminates the
above-mentioned performance problems.
Log (ZIL) blocks can be allocated from the following locations. Each
one is tried in order until the allocation succeeds:
1. dedicated log vdevs, aka "slog" (spa_log_class)
2. embedded slog metaslabs (spa_embedded_log_class)
3. other metaslabs in normal vdevs (spa_normal_class)
The space required for the embedded slog metaslabs is usually between
0.5% and 1.0% of the pool, and comes out of the existing 3.2% of "slop"
space that is not available for user data.
On an all-ssd system with 4TB storage, 87% fragmentation, 60% capacity,
and recordsize=8k, testing shows a ~50% performance increase on random
8k sync writes. On even more fragmented systems (which hit problem #3
above and call txg_wait_synced()), the performance improvement can be
arbitrarily large (>100x).
Reviewed-by: Serapheim Dimitropoulos <serapheim@delphix.com>
Reviewed-by: George Wilson <gwilson@delphix.com>
Reviewed-by: Don Brady <don.brady@delphix.com>
Reviewed-by: Mark Maybee <mark.maybee@delphix.com>
Signed-off-by: Matthew Ahrens <mahrens@delphix.com>
Closes#11389
Anton Gubarkov
Matthew Ahrens
George Melikov
Rich Ercolani
Justin Gottula
Ryan Moeller
Laurențiu Nicola
наб
Brian Behlendorf
Serapheim Dimitropoulos
grembo
Manoj Joseph
vermavipinkumar
Arshad Hussain
Toomas Soome
Yuri Pankov
Colm
pablofsf
Andrea Gelmini
Mike Swanson
Chunwei Chen
George Wilson
Tony Hutter
Thomas Lamprecht
Martin Matuška
Christian Schwarz
Andriy Gapon
José Luis Salvador Rufo
Arshad Hussain
nssrikanth
Allan Jude