I have some command which launches in this way:
./mycommand -arg=word
I measure execution time and write output to file:
/usr/bin/time -f "%K %e" ./mycommand -arg=word 2>file
However, mycommand has also stderr output. So file contains mixed data from both (time and mycommand). How can I suppress mycommand output and save time data?
If the nature of your process is such that a shell startup won't interfere with the timing too much, one simple approach is to have a shell run the direction as a subprocess of time:
/usr/bin/time -f "%K %e" sh -c '"$0" "$#" >/dev/null 2>&1' ./mycommand -arg=word
I would also suggest using the bash-builtin version of time instead of the external one, which lets you use tricks documented in BashFAQ #32:
TIMEFORMAT='%R' # replace for whatever is equivalent to your "%K %e"
timing=$(time { ./mycommand >/dev/null 2>&1; } 2>&1)
I have this line of code:
{ time cp $PWD/my_file $PWD/new_file ; } 2> my_log.log
I need to know how long it takes to execute the 'cp' command and I also need to get the PID of the 'cp'. I just want to print the PID of the 'cp' process and get the following in the my_log.log file:
<output of time>
I have tried PID=$! but this does not provide PID of the cp process.
First, you need to send your (timed) cp command to the background with a trailing &, so you can inspect the running processes after launching it.
(I suspect you're already doing this, but it's not currently reflected in the question).
$!, the special variable that contains the PID of the most recently launched background job, in this case reflects the subshell that runs the time command, so we know that it is the parent process of the cp command. To get the (one and only, in this case) child process:
If your platform has the nonstandard pgrep utility (comes with many Linux distros and BSD/macOS platforms), use:
pgrep -P $!
Otherwise, use the following POSIX-compliant approach:
ps -o pid=,ppid= | awk -v ppid=$! '$2 == ppid { print $1 }'
To put it all together, using prgep for convenience:
# Send the timed `cp` command to the background with a trailing `&`
{ time cp "$PWD/my_file" "$PWD/new_file"; } 2> my_log.log &
# Get the `cp` comand's PID via its parent PID, $!
cpPid=$(pgrep -P $!)
To know approximately how long takes cp command you can check new file size
size=$(stat -c %s "${old_file}")
cp "${old_file}" "${new_file}" &
cp_pid=$!
while kill -0 ${cp_pid}; do
cpsize=$(stat -c %s "${new_file}")
echo elapsed time $(ps -p${cp_pid} -oetime=)
echo $((100*cpsize/size)) % done so far..
sleep 3
done
EDIT: following comment stat -c %s "${file}" can be replaced by du "${file}" it's POSIX and more suitable command (see man page).
The simplest one I can think of is
pgrep cp
OK -- from the comments: "Contents of my_log.log will be PID of the cp command followed by the timing output of the cp command":
( time cp $PWD/my_file $PWD/new_file & 2>&1; echo $! ) > my_log.log 2>&1
First, you need to use /usr/bin/time explicitly, and pass the options to append to an output file. Then, use pgrep on the name of the file you are copying (cp will get too many hits):
/usr/bin/time --output=my_log.log --append cp $PWD/my_file $PWD/new_file & pgrep -f my_file > my_log.log
You may want to change the output format, because it's kinda ugly:
18400
0.00user 0.30system 0:02.43elapsed 12%CPU (0avgtext+0avgdata 2520maxresident)k
0inputs+496424outputs (0major+141minor)pagefaults 0swaps
Following what written in one of your comment above... the following is a proper answer:
The following code (just an example):
time (sleep 10 & echo -n "$!"; wait)
will return something like:
30406
real 0m10.009s
user 0m0.004s
sys 0m0.005s
In your case:
time (cp $PWD/old_file $PWD/new_file & echo -n "$!"; wait) &> my_log.log
will do the job.
I find this solution quite elegant since it's a "one liner" despite the "completely negligible overhead" you got in the timing (the timing will be related to the entire subshell (also the echo and the wait). That the overhead is negligible is evident from the result of the sleep command.
The &> redirect stdout and stderr to the same file (so you do not need to specify the 1>&2).
Pay attention that doing
(time sleep 10 & echo -n "$!")
You will get the pid of the time process not sleep or cp in your case.
I'm trying to write a script that will run and time a given and output that to a file in a .csv format.
So far from looking at SO previous posts, I've found that sh -c "$index_of_command_arg" can be used to invoke that command.
I'm also familiar with time and I know that people use /usr/bin/time for formatting, but I need to format the time given in total seconds (for example, 1.34516) but the only given option to format the real time is %E which return [hours:]minutes:seconds. Is there any way to format it the way I need?
The general idea of my script is:
# ----
# some input validation
# ----
rule=$1
command=$2
execution_time=/usr/bin/time -f "%total_seconds" sh -c "$command" #is this line possible?
echo "$rule,$execution_time" > output_file.csv
Can this be formatted the way I want? and also, the line with the comment after it,
Will this even work the way I wrote it? is the syntax correct?
Say I use the normal time and I get the real 0m2.003 ...
output, how can I take the 2.003 out of it?
The normal time you are mentioning is the bash built-in time. From the Bash Reference Manual:
The TIMEFORMAT variable may be set to a format string
that specifies how the timing information should be displayed.
…
%[p][l]R
The elapsed time in seconds.
So, you could use
execution_time=`TIMEFORMAT=%R bash -c "time $command" 2>&1 >/dev/tty`
You replace:
execution_time=/usr/bin/time -f "%total_seconds" sh -c "$command" `
with:
execution_time=`(time sh -c "$command > /dev/null 2>&1") 2>&1 | grep real |sed "s/.*m//;s/s.*//;"`
or with:
execution_time=`(time sh -c "$command > /dev/null 2>&1") 2>&1 | grep real | cut -c 8-12`
Why > /dev/null - you don't want to see $command output 2>&1 (STDERR > STDOUT)
Why 2>&1 for time command (for the same reason as in previous bullet)
I want to copy stdout to a log file from within a bash script, meaning I don't want to call the script with output piped to tee, I want the script itself to handle it. I've successfully used this answer to accomplish this, using the following code:
#!/bin/bash
exec > >(sed "s/^/[${1}] /" | tee -a myscript.log)
exec 2>&1
# <rest of script>
echo "hello"
sleep 10
echo "world"
This works, but has the downside of output being buffered until the script is completed, as is also discussed in the linked answer. In the above example, both "hello" and "world" will show up in the log only after the 10 seconds have passed.
I am aware of the stdbuf command, and if running the script with
stdbuf -oL ./myscript.sh
then stdout is indeed continuously printed both to the file and the terminal.
However, I'd like this to be handled from within the script as well. Is there any way to combine these two solutions? I'd rather not resort to a wrapper script that simply calls the original script enclosed with "stdbuf -oL".
You can use a workaround and make the script execute itself with stdbuf, if a special argument is present:
#!/bin/bash
if [[ "$1" != __BUFFERED__ ]]; then
prog="$0"
stdbuf -oL "$prog" __BUFFERED__ "$#"
else
shift #discard __BUFFERED__
exec > >(sed "s/^/[${1}] /" | tee -a myscript.log)
exec 2>&1
# <rest of script>
echo "hello"
sleep 1
echo "world"
fi
This will mostly work:
if you run the script with ./test, it shows unbuffered [] hello\n[] world.
if you run the script with ./test 123 456, it shows [123] hello\n[123] world like you want.
it won't work, however, if you run it with bash test - $0 is set to test which is not your script. Fixing this is not in the scope of this question though.
The delay in your first solution is caused by sed, not by tee. Try this instead:
#!/bin/bash
exec 6>&1 2>&1>&>(tee -a myscript.log)
To "undo" the tee effect:
exec 1>&6 2>&6 6>&-
I have a script which will be run interactively by non-technical users. The script writes status updates to STDOUT so that the user can be sure that the script is running OK.
I want both STDOUT and STDERR redirected to the terminal (so that the user can see that the script is working as well as see if there was a problem). I also want both streams redirected to a log file.
I've seen a bunch of solutions on the net. Some don't work and others are horribly complicated. I've developed a workable solution (which I'll enter as an answer), but it's kludgy.
The perfect solution would be a single line of code that could be incorporated into the beginning of any script that sends both streams to both the terminal and a log file.
EDIT: Redirecting STDERR to STDOUT and piping the result to tee works, but it depends on the users remembering to redirect and pipe the output. I want the logging to be fool-proof and automatic (which is why I'd like to be able to embed the solution into the script itself.)
Use "tee" to redirect to a file and the screen. Depending on the shell you use, you first have to redirect stderr to stdout using
./a.out 2>&1 | tee output
or
./a.out |& tee output
In csh, there is a built-in command called "script" that will capture everything that goes to the screen to a file. You start it by typing "script", then doing whatever it is you want to capture, then hit control-D to close the script file. I don't know of an equivalent for sh/bash/ksh.
Also, since you have indicated that these are your own sh scripts that you can modify, you can do the redirection internally by surrounding the whole script with braces or brackets, like
#!/bin/sh
{
... whatever you had in your script before
} 2>&1 | tee output.file
Approaching half a decade later...
I believe this is the "perfect solution" sought by the OP.
Here's a one liner you can add to the top of your Bash script:
exec > >(tee -a $HOME/logfile) 2>&1
Here's a small script demonstrating its use:
#!/usr/bin/env bash
exec > >(tee -a $HOME/logfile) 2>&1
# Test redirection of STDOUT
echo test_stdout
# Test redirection of STDERR
ls test_stderr___this_file_does_not_exist
(Note: This only works with Bash. It will not work with /bin/sh.)
Adapted from here; the original did not, from what I can tell, catch STDERR in the logfile. Fixed with a note from here.
The Pattern
the_cmd 1> >(tee stdout.txt ) 2> >(tee stderr.txt >&2 )
This redirects both stdout and stderr separately, and it sends separate copies of stdout and stderr to the caller (which might be your terminal).
In zsh, it will not proceed to the next statement until the tees have finished.
In bash, you may find that the final few lines of output appear after whatever statement comes next.
In either case, the right bits go to the right places.
Explanation
Here's a script (stored in ./example):
#! /usr/bin/env bash
the_cmd()
{
echo out;
1>&2 echo err;
}
the_cmd 1> >(tee stdout.txt ) 2> >(tee stderr.txt >&2 )
Here's a session:
$ foo=$(./example)
err
$ echo $foo
out
$ cat stdout.txt
out
$ cat stderr.txt
err
Here's how it works:
Both tee processes are started, their stdins are assigned to file descriptors. Because they're enclosed in process substitutions, the paths to those file descriptors are substituted in the calling command, so now it looks something like this:
the_cmd 1> /proc/self/fd/13 2> /proc/self/fd/14
the_cmd runs, writing stdout to the first file descriptor, and stderr to the second one.
In the bash case, once the_cmd finishes, the following statement happens immediately (if your terminal is the caller, then you will see your prompt appear).
In the zsh case, once the_cmd finishes, the shell waits for both of the tee processes to finish before moving on. More on this here.
The first tee process, which is reading from the_cmd's stdout, writes a copy of that stdout back to the caller because that's what tee does. Its outputs are not redirected, so they make it back to the caller unchanged
The second tee process has it's stdout redirected to the caller's stderr (which is good, because it's stdin is reading from the_cmd's stderr). So when it writes to its stdout, those bits go to the caller's stderr.
This keeps stderr separate from stdout both in the files and in the command's output.
If the first tee writes any errors, they'll show up in both the stderr file and in the command's stderr, if the second tee writes any errors, they'll only show up only in the terminal's stderr.
the to redirect stderr to stdout append this at your command: 2>&1
For outputting to terminal and logging into file you should use tee
Both together would look like this:
mycommand 2>&1 | tee mylogfile.log
EDIT: For embedding into your script you would do the same. So your script
#!/bin/sh
whatever1
whatever2
...
whatever3
would end up as
#!/bin/sh
( whatever1
whatever2
...
whatever3 ) 2>&1 | tee mylogfile.log
EDIT:
I see I got derailed and ended up answering a different question from the one asked. The answer to the real question is at the bottom of Paul Tomblin's answer. (If you want to enhance that solution to redirect stdout and stderr separately for some reason, you could use the technique I describe here.)
I've been wanting an answer that preserves the distinction between stdout and stderr.
Unfortunately all of the answers given so far that preserve that distinction
are race-prone: they risk programs seeing incomplete input, as I pointed out in comments.
I think I finally found an answer that preserves the distinction,
is not race prone, and isn't terribly fiddly either.
First building block: to swap stdout and stderr:
my_command 3>&1 1>&2 2>&3-
Second building block: if we wanted to filter (e.g. tee) only stderr,
we could accomplish that by swapping stdout&stderr, filtering, and then swapping back:
{ my_command 3>&1 1>&2 2>&3- | stderr_filter;} 3>&1 1>&2 2>&3-
Now the rest is easy: we can add a stdout filter, either at the beginning:
{ { my_command | stdout_filter;} 3>&1 1>&2 2>&3- | stderr_filter;} 3>&1 1>&2 2>&3-
or at the end:
{ my_command 3>&1 1>&2 2>&3- | stderr_filter;} 3>&1 1>&2 2>&3- | stdout_filter
To convince myself that both of the above commands work, I used the following:
alias my_command='{ echo "to stdout"; echo "to stderr" >&2;}'
alias stdout_filter='{ sleep 1; sed -u "s/^/teed stdout: /" | tee stdout.txt;}'
alias stderr_filter='{ sleep 2; sed -u "s/^/teed stderr: /" | tee stderr.txt;}'
Output is:
...(1 second pause)...
teed stdout: to stdout
...(another 1 second pause)...
teed stderr: to stderr
and my prompt comes back immediately after the "teed stderr: to stderr", as expected.
Footnote about zsh:
The above solution works in bash (and maybe some other shells, I'm not sure), but it doesn't work in zsh. There are two reasons it fails in zsh:
the syntax 2>&3- isn't understood by zsh; that has to be rewritten
as 2>&3 3>&-
in zsh (unlike other shells), if you redirect a file descriptor
that's already open, in some cases (I don't completely understand how it decides) it does a built-in tee-like behavior instead. To avoid this, you have to close each fd prior to
redirecting it.
So, for example, my second solution has to be rewritten for zsh as {my_command 3>&1 1>&- 1>&2 2>&- 2>&3 3>&- | stderr_filter;} 3>&1 1>&- 1>&2 2>&- 2>&3 3>&- | stdout_filter (which works in bash too, but is awfully verbose).
On the other hand, you can take advantage of zsh's mysterious built-in implicit teeing to get a much shorter solution for zsh, which doesn't run tee at all:
my_command >&1 >stdout.txt 2>&2 2>stderr.txt
(I wouldn't have guessed from the docs I found that the >&1 and 2>&2 are the thing that trigger zsh's implicit teeing; I found that out by trial-and-error.)
Use the tee program and dup stderr to stdout.
program 2>&1 | tee > logfile
Use the script command in your script (man 1 script)
Create a wrapper shellscript (2 lines) that sets up script() and then calls exit.
Part 1: wrap.sh
#!/bin/sh
script -c './realscript.sh'
exit
Part 2: realscript.sh
#!/bin/sh
echo 'Output'
Result:
~: sh wrap.sh
Script started, file is typescript
Output
Script done, file is typescript
~: cat typescript
Script started on fr. 12. des. 2008 kl. 18.07 +0100
Output
Script done on fr. 12. des. 2008 kl. 18.07 +0100
~:
I created a script called "RunScript.sh". The contents of this script is:
${APP_HOME}/${1}.sh ${2} ${3} ${4} ${5} ${6} 2>&1 | tee -a ${APP_HOME}/${1}.log
I call it like this:
./RunScript.sh ScriptToRun Param1 Param2 Param3 ...
This works, but it requires the application's scripts to be run via an external script. It's a bit kludgy.
A year later, here's an old bash script for logging anything. For example,
teelog make ... logs to a generated log name (and see the trick for logging nested makes too.)
#!/bin/bash
me=teelog
Version="2008-10-9 oct denis-bz"
Help() {
cat <<!
$me anycommand args ...
logs the output of "anycommand ..." as well as displaying it on the screen,
by running
anycommand args ... 2>&1 | tee `day`-command-args.log
That is, stdout and stderr go to both the screen, and to a log file.
(The Unix "tee" command is named after "T" pipe fittings, 1 in -> 2 out;
see http://en.wikipedia.org/wiki/Tee_(command) ).
The default log file name is made up from "command" and all the "args":
$me cmd -opt dir/file logs to `day`-cmd--opt-file.log .
To log to xx.log instead, either export log=xx.log or
$me log=xx.log cmd ...
If "logdir" is set, logs are put in that directory, which must exist.
An old xx.log is moved to /tmp/\$USER-xx.log .
The log file has a header like
# from: command args ...
# run: date pwd etc.
to show what was run; see "From" in this file.
Called as "Log" (ln -s $me Log), Log anycommand ... logs to a file:
command args ... > `day`-command-args.log
and tees stderr to both the log file and the terminal -- bash only.
Some commands that prompt for input from the console, such as a password,
don't prompt if they "| tee"; you can only type ahead, carefully.
To log all "make" s, including nested ones like
cd dir1; \$(MAKE)
cd dir2; \$(MAKE)
...
export MAKE="$me make"
!
# See also: output logging in screen(1).
exit 1
}
#-------------------------------------------------------------------------------
# bzutil.sh denisbz may2008 --
day() { # 30mar, 3mar
/bin/date +%e%h | tr '[A-Z]' '[a-z]' | tr -d ' '
}
edate() { # 19 May 2008 15:56
echo `/bin/date "+%e %h %Y %H:%M"`
}
From() { # header # from: $* # run: date pwd ...
case `uname` in Darwin )
mac=" mac `sw_vers -productVersion`"
esac
cut -c -200 <<!
${comment-#} from: $#
${comment-#} run: `edate` in $PWD `uname -n` $mac `arch`
!
# mac $PWD is pwd -L not -P real
}
# log name: day-args*.log, change this if you like --
logfilename() {
log=`day`
[[ $1 == "sudo" ]] && shift
for arg
do
log="$log-${arg##*/}" # basename
(( ${#log} >= 100 )) && break # max len 100
done
# no blanks etc in logfilename please, tr them to "-"
echo $logdir/` echo "$log".log | tr -C '.:+=[:alnum:]_\n' - `
}
#-------------------------------------------------------------------------------
case "$1" in
-v* | --v* )
echo "$0 version: $Version"
exit 1 ;;
"" | -* )
Help
esac
# scan log= etc --
while [[ $1 == [a-zA-Z_]*=* ]]; do
export "$1"
shift
done
: ${logdir=.}
[[ -w $logdir ]] || {
echo >&2 "error: $me: can't write in logdir $logdir"
exit 1
}
: ${log=` logfilename "$#" `}
[[ -f $log ]] &&
/bin/mv "$log" "/tmp/$USER-${log##*/}"
case ${0##*/} in # basename
log | Log ) # both to log, stderr to caller's stderr too --
{
From "$#"
"$#"
} > $log 2> >(tee /dev/stderr) # bash only
# see http://wooledge.org:8000/BashFAQ 47, stderr to a pipe
;;
* )
#-------------------------------------------------------------------------------
{
From "$#" # header: from ... date pwd etc.
"$#" 2>&1 # run the cmd with stderr and stdout both to the log
} | tee $log
# mac tee buffers stdout ?
esac
This question seems has not been gracefully solved yet.
Everytime I search "how to output to stdout and stderr at same time", google directs me to this post.
Today, I finally found a simple & effective way to solve almost all these kind of needs.
The essential idea is the tee command which can print to multiple output at same time, and linux-specific /proc/self/fd/{1,2,...} to represent stdout, stderr...
print stdin to both stdout and stderr
tee /proc/self/fd/2
print stdin to both stdout and stderr, and file
tee /proc/self/fd/2 file
Hope this be helpful.
Here is a solution working for bash by redirection, by combining the solution of "kvantour, MatrixManAtYrService" and "Jason Sydes":
#!/bin/bash
exec 1> >(tee x.log) 2> >(tee x.err >&2)
echo "test for log"
echo "test for err" 1>&2
Save the script above as x.sh. After running ./x.sh, the x.log only include the stdout, while the x.err only include the stderr.