Section: LIBNBD (1)
nbdcopy - copy to and from an NBD server
nbdcopy [--allocated] [-C N|--connections=N]
[--request-size=N] [-R N|--requests=N] [-S N|--sparse=N]
[--synchronous] [-T N|--threads=N] [-v|--verbose]
SOURCE, DESTINATION := - | FILE | DEVICE | NBD-URI | [ CMD ARGS ... ]
DESTINATION += null:
This copies everything from the NBD
server at "example.com"
local file called local.img
This streams the first part of the disk on the NBD
into the file
(1) command. Note here that "-"
to stream to stdout (and therefore into the pipe to the file command).
nbdinfo(1) is another way to detect the content on an NBD server.
Copy the full local hard disk "/dev/sdX"
to the NBD
on the Unix domain socket /tmp/unixsock
. Because of the -p
option this will print a progress bar.
Copy a full disk from one NBD
server to another.
(8) as a subprocess to open URL
, which is then streamed to stdout
), converted to blocks of raw format data. Note "--"
qemu-nbd flags from being interpreted as nbdcopy flags.
cat disk1 disk2 | nbdcopy --- - [ qemu-nbd -f qcow2 output.qcow2 ]
Concatenate two raw-format disk images into the qcow2 file
. The output file has to be precreated.
Read the contents of the NBD
server as fast as possible and throw it
away (write it to a virtual null device). This is useful for
servers and/or nbdcopy.
nbdcopy copies to and from an NBD
server. It can upload a local file
to an NBD
server, or download the contents of an NBD
server to a local
file, device or stdin/stdout. It can also copy between NBD
The local file can be a file, a block device (eg. "/dev/cdrom"), or
"-" which means stream in from stdin or stream out to stdout.
The NBD server can be specified using an NBD URI (like
"nbd://localhost"). The NBD server can be local or remote, and
encryption can be used if libnbd was built with encryption support.
Alternately you can use square brackets around a qemu-nbd(8) or
nbdkit(1) command to run the NBD server as a subprocess of nbdcopy.
The destination may be the special word "null:" to throw away the
For more complex copying operations including converting between disk
formats use "qemu-img convert", see qemu-img(1).
Display brief command line help and exit.
Normally nbdcopy tries to create sparse output (with holes) if the
destination supports that. It does this in two ways: either using
extent information from the source to copy holes (see
--no-extents), or by detecting runs of zeroes (see -S). If you
use --allocated then nbdcopy creates a fully allocated, non-sparse
output on the destination.
- -C N
Set the maximum number of NBD connections (``multi-conn''). By default
nbdcopy will try to use multi-conn with up to 4 connections if the NBD
server supports it. If copying between NBD servers then nbdcopy
cannot use multi-conn if either of the servers does not support it.
Assume the destination is already zeroed. This allows nbdcopy to skip
copying blocks of zeroes from the source to the destination. This is
not safe unless the destination device is already zeroed.
(--target-is-zero is provided for compatibility with
Flush writes to ensure that everything is written to persistent
storage before nbdcopy exits.
Normally nbdcopy uses extent metadata to skip over parts of the source
disk which contain holes. If you use this flag, nbdcopy ignores
extent information and reads everything, which is usually slower. You
might use this flag in two situations: the source NBD server has
incorrect metadata information; or the source has very slow extent
querying so it's faster to simply read all of the data.
Display a progress bar.
Write a progress bar to the file descriptor "FD" (a number) in a
format which is easily parsable by other programs. nbdcopy will
periodically write the string "N/100\n" (where N is an integer
between 0 and 100) to the file descriptor.
To get nbdcopy to write the progress bar to a file you can use the
following shell commands:
nbdcopy --progress=3 ...
Set the maximum request size in bytes. The maximum value is 32 MiB,
specified by the NBD protocol.
- -R N
Set the maximum number of requests in flight per NBD connection.
- -S N
Detect all zero blocks of size N (bytes) and make them sparse on the
output. You can also turn off sparse detection using -S 0.
The default is 4096 bytes.
Force synchronous copying using the libnbd(3) synchronous (``high
level'') API. This is slow but may be necessary for some broken NBD
servers which cannot handle multiple requests in flight. This mode is
also used when streaming to and from stdio, pipes and sockets.
- -T N
Use up to N threads for copying. By default this is set to the number
of processor cores available.
Note --threads=0 means autodetect and --threads=1 means use a
Verbose output. This enables debug in libnbd (see
nbd_set_debug(3)) as well as printing other useful information.
Display the package name and version and exit.
MULTI-CONN, THREADS, REQUESTS IN FLIGHT
The three options --connections
related and control the amount of parallelism available. The defaults
should ensure a reasonable amount of parallelism if possible and you
don’t need to adjust them, but this section tries to describe what is
Firstly if either side of the copy is streaming to or from stdio, a
pipe, or a socket, or if you use the --synchronous option, then
nbdcopy works in synchronous mode with no parallelism, and nothing
else in this section applies.
The --connections=N option controls NBD multi-conn (see
``Multi-conn'' in libnbd(3)), opening up to N connections to the NBD
server (or to both NBD servers if copying between NBD servers). This
defaults to 4. The NBD servers must support and advertise multi-conn.
For nbdkit(1) availability of multi-conn can depend on the plugin.
You can use nbdinfo(1) to find out if a particular NBD server is
advertising multi-conn. If the NBD server doesn’t advertise
multi-conn then only one connection will be opened regardless of the
When copying between two NBD servers, the number of connections is
limited to the minimum multi-conn supported on both sides. For the
purposes of this calculation, you can consider local files and block
devices as supporting infinite multi-conn.
When you run an NBD server as a subprocess (using the "[ ... ]"
syntax) multi-conn cannot be used.
The --threads=N option allows nbdcopy to start up to N threads
(defaulting to the number of cores). However nbdcopy cannot use more
threads than the number of NBD connections.
The --requests=N option controls the maximum number of requests in
flight on each NBD connection. This enables the NBD server to process
requests in parallel even when multi-conn isn’t available or when
using a single thread. The default is chosen to allow a reasonable
amount of parallelism without using too much memory.
Because of this parallelism, nbdcopy does not read or write blocks in
order. If for some reason you require that blocks are copied in
strict order then you must use --synchronous.
RUNNING NBD SERVER AS A SUBPROCESS
Instead of connecting to an already running server using an NBD URI,
you can run an NBD
server as a subprocess using:
nbdcopy -- [ CMD ARGS ... ] ...
This requires the server to support systemd socket activation, which
both qemu-nbd(8) and nbdkit(1) support (see also
"[" and "]" must be separate command line parameters. You will
usually need to use "--" to stop nbdcopy from misinterpreting NBD
server flags as nbdcopy flags. Both the source and destination may be
subprocesses. nbdcopy cleans up the subprocess on exit.
Some examples follow.
nbdcopy --- [ qemu-nbd -f qcow2 disk.qcow2 ] - | hexdump -C
In this example, qemu-nbd
(8) is run as a subprocess. The
subprocess opens disk.qcow2
and exposes it as NBD
nbdcopy streams this to stdout ("-"
) into the pipe which is read by
nbdcopy --- [ qemu-nbd -f qcow2 disk.qcow2 ] [ nbdkit memory 1G ]
Two subprocesses are created, qemu-nbd
(8) as the source and
(1) as the destination. The qcow2 file is converted to raw
and stored temporarily in the RAM
When nbdcopy exits both servers are killed and the RAM disk goes away,
so this command has no overall effect, but is useful for testing.
COMPARING NBD SERVER CONTENT
You can use nbdcopy, cmp
(1) and bash
(1) process substitution to
compare the content of two NBD
servers for equality:
cmp <( nbdcopy nbd://server1 - ) <( nbdcopy nbd://server2 - )
Note this tests that the content is logically equal. It does not
compare the NBD metadata such as sparseness (see nbdinfo(1)
--map option). Thus for example a run of allocated zeroes in one
server will match a hole in the other.
Richard W.M. Jones
Copyright (C) 2020 Red Hat Inc.
This library is free software; you can redistribute it and/or
modify it under the terms of the GNU
Lesser General Public
License as published by the Free Software Foundation; either
version 2 of the License, or (at your option) any later version.
This library is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
Lesser General Public License for more details.
You should have received a copy of the GNU Lesser General Public
License along with this library; if not, write to the Free Software
Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA