xref: /qemu/docs/tools/qemu-nbd.rst (revision f07a5674cf97b8473e5d06d7b1df9b51e97d553f)
1.. _qemu-nbd:
2
3=====================================
4QEMU Disk Network Block Device Server
5=====================================
6
7Synopsis
8--------
9
10**qemu-nbd** [*OPTION*]... *filename*
11
12**qemu-nbd** -L [*OPTION*]...
13
14**qemu-nbd** -d *dev*
15
16Description
17-----------
18
19Export a QEMU disk image using the NBD protocol.
20
21Other uses:
22
23- Bind a /dev/nbdX block device to a QEMU server (on Linux).
24- As a client to query exports of a remote NBD server.
25
26Options
27-------
28
29.. program:: qemu-nbd
30
31*filename* is a disk image filename, or a set of block
32driver options if :option:`--image-opts` is specified.
33
34*dev* is an NBD device.
35
36.. option:: --object type,id=ID,...
37
38  Define a new instance of the *type* object class identified by *ID*.
39  See the :manpage:`qemu(1)` manual page for full details of the properties
40  supported. The common object types that it makes sense to define are the
41  ``secret`` object, which is used to supply passwords and/or encryption
42  keys, and the ``tls-creds`` object, which is used to supply TLS
43  credentials for the ``qemu-nbd`` server or client.
44
45.. option:: -p, --port=PORT
46
47  TCP port to listen on as a server, or connect to as a client
48  (default ``10809``).
49
50.. option:: -o, --offset=OFFSET
51
52  The offset into the image.
53
54.. option:: -b, --bind=IFACE
55
56  The interface to bind to as a server, or connect to as a client
57  (default ``0.0.0.0``).
58
59.. option:: -k, --socket=PATH
60
61  Use a unix socket with path *PATH*.
62
63.. option:: --image-opts
64
65  Treat *filename* as a set of image options, instead of a plain
66  filename. If this flag is specified, the ``-f`` flag should
67  not be used, instead the :option:`format=` option should be set.
68
69.. option:: -f, --format=FMT
70
71  Force the use of the block driver for format *FMT* instead of
72  auto-detecting.
73
74.. option:: -r, --read-only
75
76  Export the disk as read-only.
77
78.. option:: -A, --allocation-depth
79
80  Expose allocation depth information via the
81  ``qemu:allocation-depth`` metadata context accessible through
82  NBD_OPT_SET_META_CONTEXT.
83
84.. option:: -B, --bitmap=NAME
85
86  If *filename* has a qcow2 persistent bitmap *NAME*, expose
87  that bitmap via the ``qemu:dirty-bitmap:NAME`` metadata context
88  accessible through NBD_OPT_SET_META_CONTEXT.
89
90.. option:: -s, --snapshot
91
92  Use *filename* as an external snapshot, create a temporary
93  file with ``backing_file=``\ *filename*, redirect the write to
94  the temporary one.
95
96.. option:: -l, --load-snapshot=SNAPSHOT_PARAM
97
98  Load an internal snapshot inside *filename* and export it
99  as an read-only device, SNAPSHOT_PARAM format is
100  ``snapshot.id=[ID],snapshot.name=[NAME]`` or ``[ID_OR_NAME]``
101
102.. option:: --cache=CACHE
103
104  The cache mode to be used with the file. Valid values are:
105  ``none``, ``writeback`` (the default), ``writethrough``,
106  ``directsync`` and ``unsafe``. See the documentation of
107  the emulator's ``-drive cache=...`` option for more info.
108
109.. option:: -n, --nocache
110
111  Equivalent to :option:`--cache=none`.
112
113.. option:: --aio=AIO
114
115  Set the asynchronous I/O mode between ``threads`` (the default),
116  ``native`` (Linux only), and ``io_uring`` (Linux 5.1+).
117
118.. option:: --discard=DISCARD
119
120  Control whether ``discard`` (also known as ``trim`` or ``unmap``)
121  requests are ignored or passed to the filesystem. *DISCARD* is one of
122  ``ignore`` (or ``off``), ``unmap`` (or ``on``).  The default is
123  ``ignore``.
124
125.. option:: --detect-zeroes=DETECT_ZEROES
126
127  Control the automatic conversion of plain zero writes by the OS to
128  driver-specific optimized zero write commands.  *DETECT_ZEROES* is one of
129  ``off``, ``on``, or ``unmap``.  ``unmap``
130  converts a zero write to an unmap operation and can only be used if
131  *DISCARD* is set to ``unmap``.  The default is ``off``.
132
133.. option:: -c, --connect=DEV
134
135  Connect *filename* to NBD device *DEV* (Linux only).
136
137.. option:: -d, --disconnect
138
139  Disconnect the device *DEV* (Linux only).
140
141.. option:: -e, --shared=NUM
142
143  Allow up to *NUM* clients to share the device (default
144  ``1``), 0 for unlimited.
145
146.. option:: -t, --persistent
147
148  Don't exit on the last connection.
149
150.. option:: -x, --export-name=NAME
151
152  Set the NBD volume export name (default of a zero-length string).
153
154.. option:: -D, --description=DESCRIPTION
155
156  Set the NBD volume export description, as a human-readable
157  string.
158
159.. option:: --handshake-limit=N
160
161  Set the timeout for a client to successfully complete its handshake
162  to N seconds (default 10), or 0 for no limit.
163
164.. option:: -L, --list
165
166  Connect as a client and list all details about the exports exposed by
167  a remote NBD server.  This enables list mode, and is incompatible
168  with options that change behavior related to a specific export (such as
169  :option:`--export-name`, :option:`--offset`, ...).
170
171.. option:: --tls-creds=ID
172
173  Enable mandatory TLS encryption for the server by setting the ID
174  of the TLS credentials object previously created with the
175  :option:`--object` option; or provide the credentials needed for
176  connecting as a client in list mode.
177
178.. option:: --tls-hostname=hostname
179
180  When validating an x509 certificate received over a TLS connection,
181  the hostname that the NBD client used to connect will be checked
182  against information in the server provided certificate. Sometimes
183  it might be required to override the hostname used to perform this
184  check. For example, if the NBD client is using a tunnel from localhost
185  to connect to the remote server, the :option:`--tls-hostname` option should
186  be used to set the officially expected hostname of the remote NBD
187  server. This can also be used if accessing NBD over a UNIX socket
188  where there is no inherent hostname available. This is only permitted
189  when acting as a NBD client with the :option:`--list` option.
190
191.. option:: --fork
192
193  Fork off the server process and exit the parent once the server is running.
194
195.. option:: --pid-file=PATH
196
197  Store the server's process ID in the given file.
198
199.. option:: --tls-authz=ID
200
201  Specify the ID of a qauthz object previously created with the
202  :option:`--object` option. This will be used to authorize connecting users
203  against their x509 distinguished name.
204
205.. option:: -v, --verbose
206
207  Display extra debugging information. This option also keeps the original
208  *STDERR* stream open if the ``qemu-nbd`` process is daemonized due to
209  other options like :option:`--fork` or :option:`-c`.
210
211.. option:: -h, --help
212
213  Display this help and exit.
214
215.. option:: -V, --version
216
217  Display version information and exit.
218
219.. option:: -T, --trace [[enable=]PATTERN][,events=FILE][,file=FILE]
220
221  .. include:: ../qemu-option-trace.rst.inc
222
223Examples
224--------
225
226Start a server listening on port 10809 that exposes only the
227guest-visible contents of a qcow2 file, with no TLS encryption, and
228with the default export name (an empty string). The command is
229one-shot, and will block until the first successful client
230disconnects:
231
232::
233
234  qemu-nbd -f qcow2 file.qcow2
235
236Start a long-running server listening with encryption on port 10810,
237and allow clients with a specific X.509 certificate to connect to
238a 1 megabyte subset of a raw file, using the export name 'subset':
239
240::
241
242  qemu-nbd \
243    --object tls-creds-x509,id=tls0,endpoint=server,dir=/path/to/qemutls \
244    --object 'authz-simple,id=auth0,identity=CN=laptop.example.com,,\
245              O=Example Org,,L=London,,ST=London,,C=GB' \
246    --tls-creds tls0 --tls-authz auth0 \
247    -t -x subset -p 10810 \
248    --image-opts driver=raw,offset=1M,size=1M,file.driver=file,file.filename=file.raw
249
250Serve a read-only copy of a guest image over a Unix socket with as
251many as 5 simultaneous readers, with a persistent process forked as a
252daemon:
253
254::
255
256  qemu-nbd --fork --persistent --shared=5 --socket=/path/to/sock \
257    --read-only --format=qcow2 file.qcow2
258
259Expose the guest-visible contents of a qcow2 file via a block device
260/dev/nbd0 (and possibly creating /dev/nbd0p1 and friends for
261partitions found within), then disconnect the device when done.
262Access to bind ``qemu-nbd`` to a /dev/nbd device generally requires root
263privileges, and may also require the execution of ``modprobe nbd``
264to enable the kernel NBD client module.  *CAUTION*: Do not use
265this method to mount filesystems from an untrusted guest image - a
266malicious guest may have prepared the image to attempt to trigger
267kernel bugs in partition probing or file system mounting.
268
269::
270
271  qemu-nbd -c /dev/nbd0 -f qcow2 file.qcow2
272  qemu-nbd -d /dev/nbd0
273
274Query a remote server to see details about what export(s) it is
275serving on port 10809, and authenticating via PSK:
276
277::
278
279  qemu-nbd \
280    --object tls-creds-psk,id=tls0,dir=/tmp/keys,username=eblake,endpoint=client \
281    --tls-creds tls0 -L -b remote.example.com
282
283See also
284--------
285
286:manpage:`qemu(1)`, :manpage:`qemu-img(1)`
287