I have a constantly running script that I output to a log file:

script.sh >> /var/log/logfile

I'd like to add a timestamp before each line that is appended to the log. Like:

Sat Sep 10 21:33:06 UTC 2011 The server has booted up.  Hmmph.

Is there any jujitsu I can use?


17 Answers 17


You can pipe the script's output through a loop that prefixes the current date and time:

./script.sh | while IFS= read -r line; do printf '%s %s\n' "$(date)" "$line"; done >>/var/log/logfile

If you'll be using this a lot, it's easy to make a bash function to handle the loop:

adddate() {
    while IFS= read -r line; do
        printf '%s %s\n' "$(date)" "$line";

./thisscript.sh | adddate >>/var/log/logfile
./thatscript.sh | adddate >>/var/log/logfile
./theotherscript.sh | adddate >>/var/log/logfile
  • 3
    @Nils it's a trick to prevent read from trimming whitespace at the beginning and and of the line. It sets IFS (bash's Internal Field Separator, basically a list of whitespace characters) to empty for the read command. Sep 15 '11 at 21:43
  • 2
    ...and -r ignores the escape-character "\". This should really work in all cases - great pice of scripting.
    – Nils
    Sep 16 '11 at 20:47
  • 8
    @Nils it's not completely bulletproof, since some implementations of echo interpret escape sequences. If you really want it not to mess with the content (other than adding dates), replace the echo command with printf "%s %s\n" "$(date)" "$line" Sep 17 '11 at 2:31
  • 4
    You might be interested on a ISO-8601 compliant date/timestamp: date -u +"%Y-%m-%dT%H:%M:%SZ" or maybe more pretty date +"%Y-%m-%d %T".
    – Pablo A
    Jul 29 '17 at 1:40
  • 2
    while this script works as expected, it spawns a new process (execute date) for each log line, which can be a massive drawback depending on your machine and the amount of logs. I would rather suggest to use ts if available, see the answer from @willem May 9 '18 at 13:20

See ts from the Ubuntu moreutils package:

command | ts

Or, if $command does automatic buffering (requires expect-dev package):

unbuffer command | ts
  • 1
    FYI, unbuffer is in the "expect" package on Arch Linux.
    – l0b0
    Jan 20 at 0:30

The date command will provide that information

date -u
Sat Sep 10 22:39:24 UTC 2011

so you can

echo $(date -u) "Some message or other"

is that what you wanted ?

  • Using the date command was kind of what I had in mind, but I can't really add that to the script itself, so what I'm looking for is a way to change this line: "script.sh >> /var/log/logfile" to append the date. Sep 10 '11 at 21:53
  • In that case redirect the output of your script to a named pipe and have a daemon listening for output which takes the script output and adds a date before writing it to a log file. You can probably modify the script I wrote here to do this. I would do it because it interests me but it's late in the UK and I have an early start tomorrow.
    – user9517
    Sep 10 '11 at 22:12

You can simply echo the command outputs to the logfile. ie,

echo "`date -u` `./script.sh`" >> /var/log/logfile

It really works :)


[sparx@E1]$ ./script.sh 
Hello Worldy
[sparx@E1]$ echo "`date -u` `./script.sh`" >> logfile.txt
[sparx@E1]$ cat logfile.txt 
Mon Sep 12 20:18:28 UTC 2011 Hello Worldy
  • What are you getting when you execute the command ?
    – SparX
    Sep 12 '11 at 20:15
  • 14
    That puts a timestamp before the entire output of ''./script.sh'', not before each line.
    – clacke
    Mar 9 '15 at 15:01

Make a config.sh file

#!/usr/bin/env bash
TIMESTAMP=`date "+%Y-%m-%d %H:%M:%S"`

When you need to send to log file use

#!/usr/bin/env bash
source /path/to/config.sh

echo "$TIMESTAMP Say what you are doing" >> $LOGFILE

do_what_you_want >> $LOGFILE

Log file will looks like

2013-02-03 18:22:30 Say what you are doing

So it will be easy to sort by date

  • 12
    Your ''config.sh'' will run ''date'' exactly once, on ''source .../config.sh''.
    – clacke
    Mar 9 '15 at 15:03

The accepted answer https://serverfault.com/a/310104 can be a bit slow, if a lot of lines have to be processed, with the overhead of starting the date process allowing about 50 lines per second in Ubuntu, and only about 10-20 in Cygwin.

When bash can be assumed a faster alternative would be the printf builtin with its %(...)T format specifier. Compare

>> while true; do date; done | uniq -c
     47 Wed Nov  9 23:17:18 STD 2016
     56 Wed Nov  9 23:17:19 STD 2016
     55 Wed Nov  9 23:17:20 STD 2016
     51 Wed Nov  9 23:17:21 STD 2016
     50 Wed Nov  9 23:17:22 STD 2016

>> while true; do printf '%(%F %T)T\n'; done | uniq -c
  20300 2016-11-09 23:17:56
  31767 2016-11-09 23:17:57
  32109 2016-11-09 23:17:58
  31036 2016-11-09 23:17:59
  30714 2016-11-09 23:18:00

Remark. On my current OpenSuse work PC (July 2021) performance is significantly up, doing 1200 lines per second with date and 265,000 lines per second with printf. Not entirely clear, how this particular performance jumped by almost two orders of magnitude compared to my laptop from 5 years ago.

  • This answer deserves more attention. Even though using external program date might now be such a performance problem on nowadays machines, when a tool has everything inside, better not complicate with more tools. Plus this answers provide a simple and locally reproducible performance measurement (on my machine I get 700-1000 per second and 150k-170k per second). More answers should be of such quality. Nov 22 '19 at 15:21
  • @StéphaneGourichon There are actually cases where it can make a huge performance difference still. E.g. I use this construct for adding timing information to stdout with someprogram | while true; do date; done | uniq -c . If a program produces hundreds of lines per second, calling date for each slows the performance down, while using the builtin does not. Though curiously, right now even the builtin produces only about 4000 lines per uniq -c for me... Could be that I tested on Linux back then.
    – kdb
    Nov 23 '19 at 21:59
  • yes we agree, no repeated external program will be faster. When performance is really important, no repeated bash construct will be even faster. GNU awk based solution may provide the best, see @ChuckCottrill's answer serverfault.com/a/835534/137665 On my machine I get 3.5x speed gain by replacing yes |head -5000000 | while true; do printf '%(%F %T)T\n'; done | uniq -c with yes |head -5000000 | gawk '{ print strftime("[%Y-%m-%d %H:%M:%S]"), $0 }' >> /var/log/logfile . Nov 25 '19 at 19:17

You mean like:

(date && script.sh) >> /var/log/logfile
  • My god, people, everyone is doing back tick replacement, named pipes, etc. Just enclose both the date command and the script in parens! The guy who has the function has a legit case if there's multi-line output and the log needs to look pretty with the date on every line, but most of these solutions are overkill that doesn't use shell semantics.
    – cjc
    Sep 12 '11 at 21:11
  • 10
    That will only add the timestamp once per execution of script.sh. The OP needs a timestamp per line. Oct 10 '12 at 14:10
  • 2
    although this doesn't answer the OP question I still found it useful info.
    – User
    Nov 2 '12 at 20:14

Try this

 date +"%Y-%m-%d %T"

Call this timestamp function in every echo command:

echo "$(timestamp): write your log here" >> /var/log/<logfile>.log
  • @ shazbot:Thanks for editing, that was a typo error, i didnt notice. Jun 15 '15 at 7:08

Short answer formatted to fit the question

script.sh | gawk '{ print strftime("[%Y-%m-%d %H:%M:%S]"), $0 }' >> /var/log/logfile


awk runs fast and is able to work as Unix pipe filter and print date by itself.

gawk '{ print strftime("[%Y-%m-%d %H:%M:%S]"), $0 }'

Let's benchmark it:

yes |head -5000000 |gawk '{ print strftime("[%Y-%m-%d %H:%M:%S]"), $0 }' |uniq -c
 461592 [2017-02-28 19:46:44] y
 488555 [2017-02-28 19:46:45] y
 491205 [2017-02-28 19:46:46] y
 498568 [2017-02-28 19:46:47] y
 502605 [2017-02-28 19:46:48] y
 494048 [2017-02-28 19:46:49] y
 493299 [2017-02-28 19:46:50] y
 498005 [2017-02-28 19:46:51] y
 502916 [2017-02-28 19:46:52] y
 495550 [2017-02-28 19:46:53] y
  73657 [2017-02-28 19:46:54] y

Additional information

Sed appears to runs much faster,

sed -e "s/^/$(date -R) /"

yes |head -5000000 |sed -e "s/^/$(date -R) /" |uniq -c
5000000 Tue, 28 Feb 2017 19:57:00 -0500 y

However, on closer inspection, set does not seem to change time,

vmstat 1 | sed -e "s/^/$(date -R) /"

Because date (which is slower by the way) gets called only once.

  • 1
    this is because it is bash evaluating "s/^/$(date -R) /" and running date once before sed. Sed is passed a static string.
    – Evan Benn
    Nov 8 '18 at 22:34
  • Plain bash: yes | head -5000000 | while read line; do echo $((SECONDS)); done | uniq -c which is much slower than gawk. ts utility has similar performance as bash loop.
    – akhan
    Sep 27 '19 at 23:43
  • Perl: yes |head -5000000 |perl -ne 'print localtime."\t".$_' |uniq -c which is slightly slower than awk.
    – akhan
    Sep 28 '19 at 0:06

Another option is to setup a function to call each time you want to output data in your code:

  echo "$(date +'%Y-%m-%d %H:%M:%S" $information} >> $logFile

Then every time in your code you want to send it to the log file call

PrintLog "Stuff you want to add..." ${LogFileVariable}

Easy peasy....

  • You're missing a close paren May 6 '20 at 17:18

The below is my log file contents

xiongyu@ubuntu:~/search_start_sh$ tail restart_scrape.log 

2017-08-25 21:10:09 scrape_yy_news_main.py got down, now I will restart it

2017-08-25 21:10:09 check_yy_news_warn.py got down, now I will restart it

2017-08-25 21:14:53 scrape_yy_news_main.py got down, now I will restart it

some of my shell contents is as below

TIMESTAMP=`date "+%Y-%m-%d %H:%M:%S"`
echo "$TIMESTAMP $search_py_file got down, now I will restart it" | tee -a $log_file 

This script print the output in terminal and also saves in log file.



    if [ $# -eq 0 ]
    then cat - | while read -r message
                echo "$(date +"[%F %T %Z] -") $message" | tee -a $MY_LOG
        echo -n "$(date +'[%F %T %Z]') - " | tee -a $MY_LOG
        echo $* | tee -a $MY_LOG

echolog "My script is starting"
whoami | echolog

Sample output:

[2017-10-29 19:46:36 UTC] - My script is starting
[2017-10-29 19:46:36 UTC] - root
  • your first date command should use single quotes, not double quotes. Mar 29 '19 at 21:00

I simply do this (buddy variable is just an example, you can put anything there):

now=$( date +%Y%m%d-%H%M-%S )
buddy=`cat /proc/buddyinfo | grep Normal`
echo "$now, $buddy" >> /home/myuser/buddyinfo-logger.log

This works perfectly.


I have noticed that using xargs with printf works well and is fast.

timestamp="$(date -Iseconds)"
ls | xargs -d '\n' -l printf '%s: %s' "$timestamp" 1>&2

I write to stderr just not to pollute the stdout.

I have made a logging utility written in bash that I use in my bash scripts:


It has more features than the example above.


At the beggining of your script add the following:

set -x
PS4='[\D{%d.%m.%y} \t]'

This will add date and time to every command and also it will write it out- eg: echo "GIT_BRANCH: $GIT_BRANCH" [11.12.20 12:50:46]echo 'GIT_BRANCH: master'

references and descriptions: Bash options: https://tldp.org/LDP/abs/html/options.html

Bash PS variables: https://linoxide.com/how-tos/change-bash-prompt-variable-ps1/#:~:text=PS1%3A%20environment%20variable%20which%20contains,long%20command%20in%20many%20lines.


Use the system logger, which is available on every linux instance, docker container, mac os, etc etc.

$ uname -a
Darwin xxx xxxx Darwin Kernel Version xxx: Thu Jan 21 00:07:06 PST 2021; root:xxx/RELEASE_X86_64 x86_64 i386 xxx Darwin
$ printf "%s\n" one two | logger -sp DEBUG 
Jul  4 00:51:42  christian[19358] <Debug>: one
Jul  4 00:51:42  christian[19358] <Debug>: two
$ docker run -it --rm debian sh -c 'printf "%s\n" one two | logger --socket-errors=off -sp WARN'
<12>Jul  3 23:08:21 root: one
<12>Jul  3 23:08:21 root: two

It's old and as fundamentally a part of linux as grep, sed, awk, etc and is written in c, which makes it a faster and better option than anything above, though ts is a better fit to your question. Do not use someone's self-roll; logging is too important and complex to leave to amateurish solution or something ad hoc.

https://man7.org/linux/man-pages/man1/logger.1.html https://github.com/karelzak/util-linux/blob/master/misc-utils/logger.c


Pipe a "sed":

script.sh | sed "s|^|$('date') :: |" >> /var/log/logfile
  • 1
    as for the other answer, this is just bash running date once. sed is not updating the time per line.
    – Evan Benn
    Nov 8 '18 at 22:35

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.