xref: /freebsd/lib/geom/journal/gjournal.8 (revision bdd1243df58e60e85101c09001d9812a789b6bc4)
1.\" Copyright (c) 2006-2009 Pawel Jakub Dawidek <pjd@FreeBSD.org>
2.\" All rights reserved.
3.\"
4.\" Redistribution and use in source and binary forms, with or without
5.\" modification, are permitted provided that the following conditions
6.\" are met:
7.\" 1. Redistributions of source code must retain the above copyright
8.\"    notice, this list of conditions and the following disclaimer.
9.\" 2. Redistributions in binary form must reproduce the above copyright
10.\"    notice, this list of conditions and the following disclaimer in the
11.\"    documentation and/or other materials provided with the distribution.
12.\"
13.\" THIS SOFTWARE IS PROVIDED BY THE AUTHORS AND CONTRIBUTORS ``AS IS'' AND
14.\" ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
15.\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
16.\" ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHORS OR CONTRIBUTORS BE LIABLE
17.\" FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
18.\" DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
19.\" OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
20.\" HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
21.\" LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
22.\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
23.\" SUCH DAMAGE.
24.\"
25.\" $FreeBSD$
26.\"
27.Dd February 17, 2009
28.Dt GJOURNAL 8
29.Os
30.Sh NAME
31.Nm gjournal
32.Nd "control utility for journaled devices"
33.Sh SYNOPSIS
34.Nm
35.Cm label
36.Op Fl cfhv
37.Op Fl s Ar jsize
38.Ar dataprov
39.Op Ar jprov
40.Nm
41.Cm stop
42.Op Fl fv
43.Ar name ...
44.Nm
45.Cm sync
46.Op Fl v
47.Nm
48.Cm clear
49.Op Fl v
50.Ar prov ...
51.Nm
52.Cm dump
53.Ar prov ...
54.Nm
55.Cm list
56.Nm
57.Cm status
58.Nm
59.Cm load
60.Nm
61.Cm unload
62.Sh DESCRIPTION
63The
64.Nm
65utility is used for journal configuration on the given GEOM provider.
66The Journal and data may be stored on the same provider or on two separate
67providers.
68This is block level journaling, not file system level journaling, which means
69everything gets logged, e.g.\& for file systems, it journals both data and
70metadata.
71The
72.Nm
73GEOM class can talk to file systems, which allows the use of
74.Nm
75for file system journaling and to keep file systems in a consistent state.
76At this time, only UFS file system is supported.
77.Pp
78To configure journaling on the UFS file system using
79.Nm ,
80one should first create a
81.Nm
82provider using the
83.Nm
84utility, then run
85.Xr newfs 8
86or
87.Xr tunefs 8
88on it with the
89.Fl J
90flag which instructs UFS to cooperate with the
91.Nm
92provider below.
93There are important differences in how journaled UFS works.
94The most important one is that
95.Xr sync 2
96and
97.Xr fsync 2
98system calls do not work as expected anymore.
99To ensure that data is stored on the data provider, the
100.Nm Cm sync
101command should be used after calling
102.Xr sync 2 .
103For the best performance possible, soft-updates should be disabled when
104.Nm
105is used.
106It is also safe and recommended to use the
107.Cm async
108.Xr mount 8
109option.
110.Pp
111When
112.Nm
113is configured on top of
114.Xr gmirror 8
115or
116.Xr graid3 8
117providers, it also keeps them in a consistent state, thus
118automatic synchronization on power failure or system crash may be disabled
119on those providers.
120.Pp
121The
122.Nm
123utility uses on-disk metadata, stored in the provider's last sector,
124to store all needed information.
125This could be a problem when an existing file system is converted to use
126.Nm .
127.Pp
128The first argument to
129.Nm
130indicates an action to be performed:
131.Bl -tag -width ".Cm status"
132.It Cm label
133Configures
134.Nm
135on the given provider(s).
136If only one provider is given, both data and journal are stored on the same
137provider.
138If two providers are given, the first one will be used as data provider and the
139second will be used as the journal provider.
140.Pp
141Additional options include:
142.Bl -tag -width ".Fl s Ar jsize"
143.It Fl c
144Checksum journal records.
145.It Fl f
146May be used to convert an existing file system to use
147.Nm ,
148but only if the journal will be configured on a separate provider and if the
149last sector in the data provider is not used by the existing file system.
150If
151.Nm
152detects that the last sector is used, it will refuse to overwrite it
153and return an error.
154This behavior may be forced by using the
155.Fl f
156flag, which will force
157.Nm
158to overwrite the last sector.
159.It Fl h
160Hardcode provider names in metadata.
161.It Fl s Ar jsize
162Specifies size of the journal if only one provider is used for both data and
163journal.
164The default is one gigabyte.
165Size should be chosen based on provider's load, and not on its size;
166recommended minimum is twice the size of the physical memory installed.
167It is not recommended to use
168.Nm
169for small file systems (e.g.: only few gigabytes big).
170.El
171.It Cm clear
172Clear metadata on the given providers.
173.It Cm stop
174Stop the given provider.
175.Pp
176Additional options include:
177.Bl -tag -width ".Fl f"
178.It Fl f
179Stop the given provider even if it is opened.
180.El
181.It Cm sync
182Trigger journal switch and enforce sending data to the data provider.
183.It Cm dump
184Dump metadata stored on the given providers.
185.It Cm list
186See
187.Xr geom 8 .
188.It Cm status
189See
190.Xr geom 8 .
191.It Cm load
192See
193.Xr geom 8 .
194.It Cm unload
195See
196.Xr geom 8 .
197.El
198.Pp
199Additional options include:
200.Bl -tag -width ".Fl v"
201.It Fl v
202Be more verbose.
203.El
204.Sh EXIT STATUS
205Exit status is 0 on success, and 1 if the command fails.
206.Sh EXAMPLES
207Create a
208.Nm
209based UFS file system and mount it:
210.Bd -literal -offset indent
211gjournal load
212gjournal label da0
213newfs -J /dev/da0.journal
214mount -o async /dev/da0.journal /mnt
215.Ed
216.Pp
217Configure journaling on an existing file system, but only if
218.Nm
219allows this (i.e., if the last sector is not already used by the file system):
220.Bd -literal -offset indent
221umount /dev/da0s1d
222gjournal label da0s1d da0s1e && \e
223    tunefs -J enable -n disable da0s1d.journal && \e
224    mount -o async /dev/da0s1d.journal /mnt || \e
225    mount /dev/da0s1d /mnt
226.Ed
227.Sh SYSCTLS
228Gjournal adds the sysctl level kern.geom.journal.
229The string and integer information available is detailed below.
230The changeable column shows whether a process with appropriate privilege may
231change the value.
232.Bl -column "accept_immediatelyXXXXXX" integerXXX -offset indent
233.It Sy "sysctl name	Type	Changeable"
234.It "debug	integer	yes"
235.It "switch_time	integer	yes"
236.It "force_switch	integer	yes"
237.It "parallel_flushes	integer	yes"
238.It "accept_immediately	integer	yes"
239.It "parallel_copies	integer	yes"
240.It "record_entries	integer	yes"
241.It "optimize	integer	yes"
242.El
243.Bl -tag -width 6n
244.It Li debug
245Setting a non-zero value enables debugging at various levels.
246Debug level 1 will record actions at a journal level, relating to journal
247switches, metadata updates, etc.
248Debug level 2 will record actions at a higher level, relating to the numbers of
249entries in journals, access requests, etc.
250Debug level 3 will record verbose detail, including insertion of I/Os to the
251journal.
252.It Li switch_time
253The maximum number of seconds a journal is allowed to remain open before
254switching to a new journal.
255.It Li force_switch
256Force a journal switch when the journal uses more than N% of the free journal
257space.
258.It Li parallel_flushes
259The number of flush I/O requests to be sent in parallel when flushing the
260journal to the data provider.
261.It Li accept_immediately
262The maximum number of I/O requests accepted at the same time.
263.It Li parallel_copies
264The number of copy I/O requests to send in parallel.
265.It Li record_entries
266The maximum number of record entries to allow in a single journal.
267.It Li optimize
268Controls whether entries in a journal will be optimized by combining overlapping
269I/Os into a single I/O and reordering the entries in a journal.
270This can be disabled by setting the sysctl to 0.
271.El
272.Ss cache
273The string and integer information available for the cache level
274is detailed below.
275The changeable column shows whether a process with appropriate
276privilege may change the value.
277.Bl -column "alloc_failuresXXXXXX" integerXXX -offset indent
278.It Sy "sysctl name	Type	Changeable"
279.It "used	integer	no"
280.It "limit	integer	yes"
281.It "divisor	integer	no"
282.It "switch	integer	yes"
283.It "misses	integer	yes"
284.It "alloc_failures	integer	yes"
285.El
286.Bl -tag -width 6n
287.It Li used
288The number of bytes currently allocated to the cache.
289.It Li limit
290The maximum number of bytes to be allocated to the cache.
291.It Li divisor
292Sets the cache size to be used as a proportion of kmem_size.
293A value of 2 (the default) will cause the cache size to be set to 1/2 of the
294kmem_size.
295.It Li switch
296Force a journal switch when this percentage of cache has been used.
297.It Li misses
298The number of cache misses, when data has been read, but was not found in the
299cache.
300.It Li alloc_failures
301The number of times memory failed to be allocated to the cache because the cache
302limit was hit.
303.El
304.Ss stats
305The string and integer information available for the statistics level
306is detailed below.
307The changeable column shows whether a process with appropriate
308privilege may change the value.
309.Bl -column "skipped_bytesXXXXXX" integerXXX -offset indent
310.It Sy "sysctl name	Type	Changeable"
311.It "skipped_bytes	integer	yes"
312.It "combined_ios	integer	yes"
313.It "switches	integer	yes"
314.It "wait_for_copy	integer	yes"
315.It "journal_full	integer	yes"
316.It "low_mem	integer	yes"
317.El
318.Bl -tag -width 6n
319.It Li skipped_bytes
320The number of bytes skipped.
321.It Li combined_ios
322The number of I/Os which were combined by journal optimization.
323.It Li switches
324The number of journal switches.
325.It Li wait_for_copy
326The number of times the journal switch process had to wait for the previous
327journal copy to complete.
328.It Li journal_full
329The number of times the journal was almost full, forcing a journal switch.
330.It Li low_mem
331The number of times the low_mem hook was called.
332.El
333.Sh SEE ALSO
334.Xr geom 4 ,
335.Xr geom 8 ,
336.Xr mount 8 ,
337.Xr newfs 8 ,
338.Xr tunefs 8 ,
339.Xr umount 8
340.Sh HISTORY
341The
342.Nm
343utility appeared in
344.Fx 7.0 .
345.Sh AUTHORS
346.An Pawel Jakub Dawidek Aq Mt pjd@FreeBSD.org
347