pv - monitor the progress of data through a pipe
pv [OPTION] [FILE]...
pv [-h|-V]
pv shows the progress of data through a pipeline by giving
information such as time elapsed, percentage completed (with progress bar),
current throughput rate, total data transferred, and ETA.
To use it, insert it in a pipeline between two processes, with the
appropriate options. Its standard input will be passed through to its
standard output and progress will be shown on standard error.
pv will copy each supplied FILE in turn to standard
output (- means standard input), or if no FILEs are
specified just standard input is copied. This is the same behaviour as
cat(1).
A simple example to watch how quickly a file is transferred using
nc(1):
pv file | nc -w 1 somewhere.com 3000
A similar example, transferring a file from another process and
passing the expected size to pv:
cat file | pv -s 12345 | nc -w 1 somewhere.com
3000
A more complicated example using numeric output to feed into the
dialog(1) program for a full-screen progress display:
(tar cf - . \
| pv -n -s $(du -sb . | awk '{print $1}') \
| gzip -9 > out.tgz) 2>&1 \
| dialog --gauge 'Progress' 7 70
Taking an image of a disk, skipping errors:
pv -EE /dev/your/disk/device >
disk-image.img
Writing an image back to a disk:
pv disk-image.img > /dev/your/disk/device
Zeroing a disk:
pv < /dev/zero > /dev/your/disk/device
Note that if the input size cannot be calculated, and the output
is a block device, then the size of the block device will be used and
pv will automatically stop at that size as if -S had been
given.
(Linux only): Watching file descriptor 3 opened by another process
1234:
pv -d 1234:3
(Linux only): Watching all file descriptors used by process
1234:
pv -d 1234
pv takes many options, which are divided into display
switches, output modifiers, and general options.
If no display switches are specified, pv behaves as if
-p, -t, -e, -r, and -b had been given
(i.e. everything except average rate is switched on). Otherwise, only those
display types that are explicitly switched on will be shown.
- -p, --progress
- Turn the progress bar on. If any inputs are not files, or are unreadable,
and no size was explicitly given (with the -s modifier), the
progress bar cannot indicate how close to completion the transfer is, so
it will just move left and right to indicate that data is moving.
- -t, --timer
- Turn the timer on. This will display the total elapsed time that pv
has been running for.
- -e, --eta
- Turn the ETA timer on. This will attempt to guess, based on current
transfer rates and the total data size, how long it will be before
completion. This option will have no effect if the total data size cannot
be determined.
- -I, --fineta
- Turn the ETA timer on, but display the estimated local time of arrival
instead of time left. When the estimated time is more than 6 hours in the
future, the date is shown as well.
- -r, --rate
- Turn the rate counter on. This will display the current rate of data
transfer.
- -a, --average-rate
- Turn the average rate counter on. This will display the current average
rate of data transfer (default: last 30s, see -m).
- -b, --bytes
- Turn the total byte counter on. This will display the total amount of data
transferred so far.
- -8, --bits
- Display the total bits instead of the total bytes. The output suffix will
be "b" instead of "B".
- -T,
--buffer-percent
- Turn on the transfer buffer percentage display. This will show the
percentage of the transfer buffer in use - but see the caveat under
%T in the FORMATTING section below. Implies -C.
- -A NUM, --last-written NUM
- Show the last NUM bytes written - but see the caveat under
%nA in the FORMATTING section below. Implies -C.
- -F FORMAT, --format FORMAT
- Ignore the options -p, -t, -e, -r, -a,
-b, -T, and -A, and instead use the format string
FORMAT to determine the output format. See the FORMATTING
section below.
- -n, --numeric
- Numeric output. Instead of giving a visual indication of progress,
pv will give an integer percentage, one per line, on standard
error, suitable for piping (via convoluted redirection) into
dialog(1). Note that -f is not required if -n is
being used.
-
- Note that if --numeric is in use, then adding --bytes will
cause the number of bytes processed so far to be output instead of a
percentage; if --line-mode is also in use as well as --bytes
and --numeric, then instead of bytes or a percentage, the number of
lines so far is output. And finally, if --timer is added to
--numeric, then each output line is prefixed with the elapsed time
so far, as a decimal number of seconds.
- -q, --quiet
- No output. Useful if the -L option is being used on its own to just
limit the transfer rate of a pipe.
- -W, --wait
- Wait until the first byte has been transferred before showing any progress
information or calculating any ETAs. Useful if the program you are piping
to or from requires extra information before it starts, eg piping data
into gpg(1) or mcrypt(1) which require a passphrase before
data can be processed.
- -D SEC, --delay-start SEC
- Wait until SEC seconds have passed before showing any progress
information, for example in a script where you only want to show a
progress bar if it starts taking a long time. Note that this can be a
decimal such as 0.5.
- -s SIZE, --size SIZE
- Assume the total amount of data to be transferred is SIZE bytes
when calculating percentages and ETAs. The same suffixes of "k",
"m" etc can be used as with -L.
-
- If SIZE starts with @, the size of file whose name follows
the @ will be used.
-
- Note that --size has no effect if used with
-d PID to watch all file descriptors of a process,
but will work with -d PID:FD.
- -l, --line-mode
- Instead of counting bytes, count lines (newline characters). The progress
bar will only move when a new line is found, and the value passed to the
-s option will be interpreted as a line count.
-
- If this option is used without -s, the "total size" (in
this case, total line count) is calculated by reading through all input
files once before transfer starts. If any inputs are pipes or non-regular
files, or are unreadable, the total size will not be calculated.
- -0, --null
- Count lines as terminated with a zero byte instead of with a newline. This
option implies --line-mode.
- -i SEC, --interval SEC
- Wait SEC seconds between updates. The default is to update every
second. Note that this can be a decimal such as 0.1.
- -m SEC, --average-rate-window SEC
- Compute current average rate over a SEC seconds window for average
rate and ETA calculations (default 30 seconds).
- -w WIDTH, --width WIDTH
- Assume the terminal is WIDTH characters wide, instead of trying to
work it out (or assuming 80 if it cannot be guessed). If this option is
used, the output width will not be adjusted if the width of the terminal
changes while the transfer is running.
- -H HEIGHT, --height HEIGHT
- Assume the terminal is HEIGHT rows high, instead of trying to work
it out (or assuming 25 if it cannot be guessed). If this option is used,
the output height will not be adjusted if the height of the terminal
changes while the transfer is running.
- -N NAME, --name NAME
- Prefix the output information with NAME. Useful in conjunction with
-c if you have a complicated pipeline and you want to be able to
tell different parts of it apart.
- -f, --force
- Force output. Normally, pv will not output any visual display if
standard error is not a terminal. This option forces it to do so.
- -c, --cursor
- Use cursor positioning escape sequences instead of just using carriage
returns. This is useful in conjunction with -N (name) if you are
using multiple pv invocations in a single, long, pipeline.
- -L RATE, --rate-limit RATE
- Limit the transfer to a maximum of RATE bytes per second. A suffix
of "K", "M", "G", or "T" can be
added to denote kibibytes (*1024), mebibytes, and so on.
- -B BYTES, --buffer-size BYTES
- Use a transfer buffer size of BYTES bytes. A suffix of
"K", "M", "G", or "T" can be added
to denote kibibytes (*1024), mebibytes, and so on. The default buffer size
is the block size of the input file's filesystem multiplied by 32 (512KiB
max), or 400KiB if the block size cannot be determined. This can be useful
on platforms like MacOS which perform better in pipelines with specific
buffer sizes such as 1024. Implies -C.
- -C, --no-splice
- Never use splice(2), even if it would normally be possible. The
splice(2) system call is a more efficient way of transferring data
from or to a pipe than regular read(2) and write(2), but
means that the transfer buffer may not be used. This prevents -A
and -T from working, cannot work with -X, and makes
-B redundant, so using -A, -T, -X, or
-B automatically switches on -C. Switching on -C
results in a small loss of transfer efficiency. (This option has no effect
on systems where splice(2) is unavailable).
- -E, --skip-errors
- Ignore read errors by attempting to skip past the offending sections. The
corresponding parts of the output will be null bytes. At first only a few
bytes will be skipped, but if there are many errors in a row then the
skips will move up to chunks of 512. This is intended to be similar to
dd conv=sync,noerror but has not been as thoroughly tested.
-
- Specify -E twice to only report a read error once per file, instead
of reporting each byte range skipped.
- -Z BYTES, --error-skip-block BYTES
- When ignoring read errors with -E, instead of trying to adaptively
skip by reading small amounts and skipping progressively larger sections
until a read succeeds, move to the next file block of BYTES bytes
as soon as an error occurs. There may still be some shorter skips where
the block being skipped coincides with the end of the transfer
buffer.
-
- This option can only be used with -E and is intended for use when
reading from a block device, such as -E -Z 4K to skip
in 4 kibibyte blocks. This will speed up reads from faulty media, at the
expense of potentially losing more data.
- -S, --stop-at-size
- If a size was specified with -s, stop transferring data once that
many bytes have been written, instead of continuing to the end of
input.
- -Y, --sync
- After every write operation, synchronise the buffer caches to disk - see
fdatasync(2). This has no effect when the output is a pipe. Using
-Y may improve the accuracy of the progress bar when writing to a
slow disk.
- -K, --direct-io
- Set the O_DIRECT flag on all inputs and outputs, if it is
available. This will minimise the effect of caches, at the cost of
performance. Due to memory alignment requirements, it also may cause read
or write failures with an error of "Invalid argument",
especially if reading and writing files across a variety of filesystems in
a single pv call. Use this option with caution.
- -X, --discard
- Instead of transferring input data to standard output, discard it. This is
equivalent to redirecting standard output to /dev/null, except that
write(2) is never called. Implies -C.
- -d PID[:FD], --watchfd PID[:FD]
- Instead of transferring data, watch file descriptor FD of process
PID, and show its progress. The pv process will exit when
FD either changes to a different file, changes read/write mode, or
is closed; other data transfer modifiers - and remote control - may not be
used with this option.
-
- If only a PID is specified, then that process will be watched, and
all regular files and block devices it opens will be shown with a progress
bar. The pv process will exit when process PID exits.
- -R PID, --remote PID
- If PID is an instance of pv that is already running,
-R PID will cause that instance to act as though it
had been given this instance's command line instead. For example, if pv
-L 123K is running with process ID 9876, then running pv -R 9876 -L
321K will cause it to start using a rate limit of 321KiB instead of
123KiB. Note that some options cannot be changed while running, such as
-c, -l, -f, -D, -E, and -S.
- -P FILE, --pidfile FILE
- Save the process ID of pv in FILE. The file will be replaced
if it already exists, and will be removed when pv exits. While
pv is running, it will contain a single number - the process ID of
pv - followed by a newline.
- -h, --help
- Print a usage message on standard output and exit successfully.
- -V, --version
- Print version information on standard output and exit successfully.
If the -F option is given, then the output format is
determined by the given format string. Within that string, the following
sequences can be used:
- %p
- Progress bar. Expands to fill the remaining space. Should only be
specified once. Equivalent to -p.
- %t
- Elapsed time. Equivalent to -t.
- %e
- ETA as time remaining. Equivalent to -e.
- %I
- ETA as local time of completion. Equivalent to -I.
- %r
- Current data transfer rate. Equivalent to -r.
- %a
- Average data transfer rate. Equivalent to -a.
- %b
- Bytes transferred so far (or lines if -l was specified). Equivalent
to -b. If --bits was specified, %b shows the bits
transferred so far, not bytes.
- %T
- Percentage of the transfer buffer in use. Equivalent to -T. Shows
"{----}" if the transfer is being done with splice(2),
since splicing to or from pipes does not use the buffer.
- %nA
- Show the last n bytes written (e.g. %16A for the last 16
bytes). Shows only dots if the transfer is being done with
splice(2), since splicing to or from pipes does not use the
buffer.
- %N
- Name prefix given by -N. Padded to 9 characters with spaces, and
suffixed with :.
- %%
- A single %.
The format string equivalent of turning on all display switches is
`%N %b %T %t %r %a %p %e'.
Some suggested common switch combinations:
- pv -ptebar
- Show a progress bar, elapsed time, estimated completion time, byte
counter, average rate, and current rate.
- pv -betlap
- Show a progress bar, elapsed time, estimated completion time, line
counter, and average rate, counting lines instead of bytes.
- pv -t
- Show only the elapsed time - useful as a simple timer, e.g. sleep 10m |
pv -t.
- pv -pterb
- The default behaviour: progress bar, elapsed time, estimated completion
time, current rate, and byte counter.
On MacOS, it may be useful to specify -B 1024 in a
pipeline, as this may improve performance.
An exit status of 1 indicates a problem with the -R or
-P options.
Any other exit status is a bitmask of the following:
- 2
- One or more files could not be accessed, stat(2)ed, or opened.
- 4
- An input file was the same as the output file.
- 8
- Internal error with closing a file or moving to the next file.
- 16
- There was an error while transferring data from one or more input
files.
- 32
- A signal was caught that caused an early exit.
- 64
- Memory allocation failed.
A zero exit status indicates no problems.
Written by Andrew Wood, with patches submitted by various other
people. Please see the package's ACKNOWLEDGEMENTS file for a complete list
of contributors.
The following problems are known to exist in pv:
- In some versions of bash(1) and zsh(1), the construct
<(pv filename) will not output any progress to the terminal when
run from an interactive shell, due to the subprocess being run in a
separate process group from the one that owns the terminal. In these
cases, use --force.
- The -c option does not work properly on Cygwin without
cygserver running, if started near the bottom of the screen (IPC is
needed to handle the terminal scrolling). To fix this, start
cygserver before using pv -c.
- The -R option requires that either /run/user/<uid>/ or
$HOME/ can be written to, for inter-process communication.
If you find any other problems, please report them.
Please report any bugs to pv@ivarch.com.
Alternatively, use the issue tracker linked from the pv
home page: <https://www.ivarch.com/programs/pv.shtml>
cat(1), dialog(1), splice(2), open(2)
(for O_DIRECT)
Copyright © 2002-2008, 2010, 2012-2015, 2017, 2021, 2023
Andrew Wood.
License GPLv3+: GNU GPL version 3 or later
<https://www.gnu.org/licenses/gpl-3.0.html>.
This is free software: you are free to change and redistribute it.
There is NO WARRANTY, to the extent permitted by law.