How can I programmatically create a new cron job?

Question

I want to be able to programatically add a new cron job, what is the best way to do this?

From my research, it seems I could dump the current crontab and then append a new one, piping that back into crontab:

(crontab -l ; echo "0 * * * * wget -O - -q http://www.example.com/cron.php") | crontab -

Is there a better way?

On Solaris just remove the dash for the last crontab. You can add a grep to avoid adding a line already there. — lacroix1547, May 31 '12 at 16:46
Possible duplicate of [How to create a cron job using Bash](https://stackoverflow.com/questions/878600/how-to-create-a-cron-job-using-bash) — Ciro Santilli新疆棉花TRUMP BAN BAD, Jul 30 '17 at 07:12
curly braces instead of parentheses will do it without spawning a process. Make sure to keep spaces around the braces, { ... ; }. — Jay Winston, Apr 04 '19 at 08:18
@davidmytton - would you consider changing your selected answer to that from MarkR - i think the consensus is that it is the better answer to the question and will improve this stackoverflow entry for others — Reinsbrain, Apr 13 '21 at 16:46

score 144 · Accepted Answer · edited Jun 05 '16 at 21:48

144

The best way if you're running as root, is to drop a file into /etc/cron.d

if you use a package manager to package your software, you can simply lay down files in that directory and they are interpreted as if they were crontabs, but with an extra field for the username, e.g.:

Filename: /etc/cron.d/per_minute

Content: * * * * * root /bin/sh /home/root/script.sh

edited Jun 05 '16 at 21:48

electronix384128

6,286
10
39
64

answered Mar 04 '09 at 14:38

MarkR

59,334
14
109
144

5

Just make sure the version of cron in use supports /etc/cron.d/. Most modern Linux distributions do. – sleske Mar 06 '09 at 11:09
I think cron.d is a Redhat patch. It is however, a very widespread one, due the the fact that it's so useful for packaging. – MarkR Mar 24 '11 at 13:47
8

This is the simplest, most elegant solution. Sure beats having to parse and edit an existing crontab with other non-related entries. – Cerin Oct 07 '11 at 14:01
When your cron supports this, this is the best solution imho – Jaap Versteegh May 09 '13 at 10:40
13

One potential caveat to this is the frequency at which cron picks up new additions to this folder (once per hour). If you're expecting your job to begin running right away, beware. – JonathanK Apr 17 '14 at 02:53
6

@JonathanK, do you happen to know if cron can be asked to re-scan `/etc/cron.d/*`? It's hard to know if something is working if one has to wait for an hour to check! – halfer Dec 09 '14 at 15:57
1

At least on CentOS & RHEL, "The cron daemon checks the /etc/crontab file, the /etc/cron.d/ directory, and the /var/spool/cron/ directory every minute for any changes." https://www.centos.org/docs/5/html/5.2/Deployment_Guide/s2-autotasks-cron-configuring.html https://access.redhat.com/documentation/en-US/Red_Hat_Enterprise_Linux/5/html/Deployment_Guide/ch-autotasks.html – dskrvk Feb 10 '17 at 00:33
4

how should perms set for this file? chmod +x /etc/cron.d/per_minute ? – giò Sep 03 '17 at 09:03
This solution is very dependent on the specifics of your distribution. Not all *nix-es are compatible with `cron.d` directory and each has its own little quirks (some allow dashes in the filename, some don't, some need specific file permissions, some don't, some allow `root` crontabs, some don't, etc etc). I couldn't get it to work on Ubuntu 16.04 – Ciprian Tomoiagă Jul 12 '19 at 08:44
@giò the debian man page in Buster at least says: **The files under /etc/cron.d do not need to be executable** https://manpages.debian.org/buster/cron/cron.8.en.html – Badr Elmers Feb 10 '21 at 08:26
debian too will check the cron files every minute: from man page: **cron then wakes up every minute, examining all stored crontabs, checking each command to see if it should be run in the current minute.** – Badr Elmers Feb 10 '21 at 08:27
i'd much prefer this were the selected answer – Reinsbrain Apr 13 '21 at 16:43

score 108 · Answer 2 · edited Apr 17 '18 at 20:31

108

OP's solution has a bug, it might allow entries to be added twice, use below to fix.

(crontab -l ; echo "0 * * * * your_command") | sort - | uniq - | crontab -

edited Apr 17 '18 at 20:31

rogerdpack

50,731
31
212
332

answered Nov 13 '12 at 04:49

JohnZ

1,217
1
9
5

1

Would be helpful if you described the bug. – Cody Poll Apr 12 '13 at 15:55
5

This is actually a better if not described solution. It ensures the command is not added twice to crontab. – Bufke Apr 19 '13 at 19:17
2

This is hardly a good fix for the duplicate entry issue. What if another line was added to crontab in the meantime? – hmn Jul 12 '13 at 13:55
6

To ensure uniqueness, just `sort` before `uniq`, e.g.: `(crontab -l ; echo "0 * * * * your_command") | sort - | uniq - | crontab -` – mway Mar 21 '14 at 15:35
6

No need for `uniq`. Use the `-u` option on `sort`. – papiro Dec 12 '16 at 17:40
2

This has a (small but dangerous) race condition. – ron rothman Dec 16 '16 at 23:38
@ron.rothman would you mind expanding on the race condition ? – Arnaud P May 15 '17 at 14:29
@ArnaudP Because it does read/modify/write, there is the (slim) possibility that someone else's changes to crontab get lost. – ron rothman May 15 '17 at 15:57
7

how to avoid **no crontab for user** when I first edit crontab. – wyx Jul 22 '17 at 13:59
1

@wyx see my answer here for the `no crontab` issue :) – rogerdpack Apr 17 '18 at 20:17
@wyx the "no crontab for user" is actually from stderr, you an use `>/dev/null` to suppress it. Then the correct command should be `(crontab -l 2>/dev/null | grep -v '^[a-zA-Z]'; echo "0 * * * * your_command") | sort - | uniq - | crontab -` – Napoleon Feb 29 '20 at 12:01

score 25 · Answer 3 · edited Jul 22 '14 at 16:28

25

To Add something to cron

(crontab -l ; echo "0 * * * * hupChannel.sh") 2>&1 | grep -v "no crontab" | sort | uniq | crontab -

To remove this from cron

(crontab -l ; echo "0 * * * * hupChannel.sh") 2>&1 | grep -v "no crontab" | grep -v hupChannel.sh |  sort | uniq | crontab -

hope would help someone

edited Jul 22 '14 at 16:28

Matas Vaitkevicius

49,230
25
212
228

answered Jul 22 '14 at 16:14

user3865482

251
3
3

1

This is exactly what I was looking for I would just add this: | grep -v '^#' | to filter out the comments – Tjunkie Aug 29 '14 at 17:01
4

You actually shouldn't need `2>&1 | grep -v "no crontab"` because when there is no crontab, the output line `crontab: no crontab for...` is sent to stderr. There's no reason to capture that output, send it to stdout, and then filter it out using grep. If your goal is to avoid seeing `crontab: no crontab for...` in your output, then use `2> /dev/null | sort....`. – matty Sep 02 '16 at 13:43

Eldamir · Answer 4 · 2018-06-14T05:54:56.423

Most of the solutions here are for adding lines to the crontab. If you need more control, you'll want to be able to control the entire contents of the crontab.

You can use piping to do this pretty elegantly.

To completely rewrite the crontab, do

echo "2 2 2 2 2 /bin/echo foobar" |crontab -

This should be easy to combine with other answers described here like

crontab -l | <something> | tee |crontab -

Or, if you have the contents in a file, it is even simpler

cat <file> |crontab -

score 6 · Answer 5 · edited Dec 21 '18 at 08:08

6

Simply change the editor to tee command:

export EDITOR="tee"
echo "0 * * * * /bin/echo 'Hello World'" | crontab -e

edited Dec 21 '18 at 08:08

Destroyica

4,007
2
30
48

answered Jun 17 '14 at 14:15

Renato Tambellini

76
1
1

5

Beware - this wipes any existing entries. – starfry Jan 28 '19 at 17:17

score 6 · Answer 6 · answered Mar 04 '09 at 14:44

6

If you're planning on doing it for a run-once scenario for just wget'ing something, take a look at 'at'

answered Mar 04 '09 at 14:44

Christian Witts

10,417
1
29
42

score 5 · Answer 7 · edited Oct 09 '17 at 21:41

Assuming that there is already an entry in your crontab, the following command should work relatively well. Note that the $CMD variable is only there for readability. Sorting before filtering duplicates is important, because uniq only works on adjacent lines.

CMD='wget -O - -q http://www.example.com/cron.php"'
(crontab -l ; echo "0 * * * * $CMD") | sort | uniq | crontab -

If you currently have an empty crontab, you will receive the following error to stderr:

no crontab for user

If you want to avoid this, you can add a little bit of complexity add do something like this:

(crontab -l ; echo "0 * * * * $CMD") 2>&1 | sed "s/no crontab for $(whoami)//"  | sort | uniq | crontab -

rogerdpack · Answer 8 · 2018-04-18T15:13:56.957

Here's another one-liner way, that avoids duplicates

(crontab -l 2>/dev/null | fgrep -v "*/1 *  *  *  * your_command"; echo "*/1 *  *  *  * your_command") | crontab -

And here's a way to do JohnZ's answer and avoid no crontab for user message, or if you need to operate in a set -eu type environment and can't have anything return a failure (in which case the 2>/dev/null part is optional):

( (crontab -l 2>/dev/null || echo "")  ; echo "0 * * * * your_command") | sort -u - | crontab -

Or if you want to split things up so that they're more readable:

new_job="0 * * * * your_command"
preceding_cron_jobs=$(crontab -l || echo "")
(echo "$preceding_cron_jobs" ; echo "$new_job") | sort - | uniq - | crontab -

Or optionally remove any references to your_command (ex: if the schedule has changed, you only want it ever cron'ed once). In this case we no longer need uniq (added bonus, insertion order is also preserved):

new_job="0 * * * * your_command"
preceding_cron_jobs=$(crontab -l || echo "")
preceding_cron_jobs=$(echo "$preceding_cron_jobs" | grep -v your_command )
(echo "$preceding_cron_jobs" ; echo "$new_job") | crontab -

score 2 · Answer 9 · answered Jul 24 '15 at 16:00

2

Adding to JohnZ's answer, here's the syntax to schedule as root if you are a sudoer:

(sudo crontab -l ; echo "0 * * * * your_command") | sort - | uniq - | sudo crontab -

answered Jul 24 '15 at 16:00

BuvinJ

7,867
3
61
78

score 2 · Answer 10 · answered Mar 06 '09 at 10:41

2

man crontab is also useful:

CRONTAB(1)

NAME

   crontab - manipulate per-user crontabs (Dillon's Cron)

SYNOPSIS

   crontab file [-u user] - replace crontab from file

   crontab - [-u user] - replace crontab from stdin

   crontab -l [user] - list crontab for user

answered Mar 06 '09 at 10:41

vitaly.v.ch

2,278
3
24
35

3

The part that says "replace crontab from stdin" is actually half an answer :-) – Grodriguez Feb 13 '14 at 17:46

score 1 · Answer 11 · answered Jul 14 '14 at 00:32

function cronjob_exists($command){

    $cronjob_exists=false;

    exec('crontab -l', $crontab);


    if(isset($crontab)&&is_array($crontab)){

        $crontab = array_flip($crontab);

        if(isset($crontab[$command])){

            $cronjob_exists=true;

        }

    }
    return $cronjob_exists;
}

function append_cronjob($command){

    if(is_string($command)&&!empty($command)&&cronjob_exists($command)===FALSE){

        //add job to crontab
        exec('echo -e "`crontab -l`\n'.$command.'" | crontab -', $output);


    }

    return $output;
}

    append_cronjob('* * * * * curl -s http://localhost/cron/test.php');

score 1 · Answer 12 · answered Nov 11 '16 at 11:43

1

This would check to ensure that your command doesn't already exist before adding it.

crontab -l 2>/dev/null | grep -q '/path/to/script' || echo "5 * * * * /path/to/script" | crontab -

Cheers.

answered Nov 11 '16 at 11:43

Ezra Obiwale

1,430
11
13

score 0 · Answer 13 · answered Aug 19 '17 at 10:33

0

Piping stdout into crontab didn't install the new crontab for me on macOS, so I found this solution instead, using the tee editor in a sub shell:

(EDITOR=tee && (crontab -l ; echo "@daily ~/my-script.sh" ) | uniq - | crontab -e)

answered Aug 19 '17 at 10:33

Simon Hänisch

3,634
1
27
35

ajcg · Answer 14 · 2019-10-21T15:09:54.037

0

If you want the task to run as a user:

crontab -l | { cat; echo "@weekly what_you_want_to_execute"; } | crontab -

If you want the task to run with privileges:

sudo crontab -l | { cat; echo "@weekly what_you_want_to_execute"; } | sudo crontab -

and check task (with or without 'sudo'):

crontab -l | sed '/^$/d; /#/d'

edited Oct 21 '19 at 15:09

answered Oct 21 '19 at 14:29

ajcg

286
2
11

score 0 · Answer 15 · answered Mar 06 '09 at 10:39

0

also you can add your tasks to /etc/cron.*/

answered Mar 06 '09 at 10:39

vitaly.v.ch

2,278
3
24
35

score 0 · Answer 16 · answered Sep 23 '20 at 09:42

Below is what I use in my script

1.

(2>/dev/null crontab -l ; echo "0 3 * * * /usr/local/bin/certbot-auto renew") | crontab -

cat <(crontab -l 2>/dev/null) <(echo "0 3 * * * /usr/local/bin/certbot-auto renew") | crontab -

#write out current crontab

crontab -l > mycron 2>/dev/null

#echo new cron into cron file

echo "0 3 * * * /usr/local/bin/certbot-auto renew" >> mycron

#install new cron file

crontab mycron

rm mycron

score 0 · Answer 17 · answered Dec 05 '20 at 21:08

To have flexibility when changing the command in the future you can do this (works best with an update script)

MARK=SOME_UNIQUE_MARK_TEXT
LINE="This is the command # $MARK"
# NOTE: I'm using -e because I might want to avoid weird bash expansions for '*' or '$' and place:
# \x2A instead of *
# \x24 instead of $
( crontab -l | grep -v $MARK ; echo -e "0 * * * *" $LINE ) | crontab -

In this way I can update an old crontab command which no longer serves current purpose.

Basically I get the current crontab, I eliminate the line containing the MARK (grep -v) and replace it with the new one by appending it at the end of the crontab file via echo -e. In this particular case I want it executed every hour at minute 0 0 * * * * as it can be seen from here.

score -2 · Answer 18 · answered Mar 04 '09 at 14:36

-2

You could also edit the cron table text file directly, but your solution seems perfectly acceptable.

answered Mar 04 '09 at 14:36

Alex Fort

17,293
5
40
51

1

I would be leery of this due to concerns of the chance of concurrent edits causing file corruption. Using the command-line crontab command should avoid that problem. – Craig S Mar 05 '09 at 02:28
Craig, without more research I wouldn't be sure that the command line version is atomic and race-condition-safe. Probably "quite safe" anyway. – tuomassalo Mar 16 '11 at 13:03
"Editing the file directly" is not feasible without root access (provided you can figure out where the user's crontab file lives, and how to make sure the cron daemon correctly receives your changes), and "editing" programmatically seems to be what the question wants advice on in the first place. – tripleee May 19 '15 at 09:12

score -27 · Answer 19 · answered Mar 04 '09 at 14:38

-27

It's always worked well for me.

You should consider a slightly more sophisticated script that can do three things.

Append a crontab line; assuring that it didn't exist. Adding when it already exists is bad.
Remove the crontab line. Perhaps only warning if it didn't exist.
A combination of the above two features to replace the crontab line.

answered Mar 04 '09 at 14:38

S.Lott

359,791
75
487
757

80

He asks how and you tell him what? – Cerin Oct 07 '11 at 14:01
1

What if the user does not have a crontab yet? – Maxim Egorushkin Jun 29 '16 at 15:48
Please provide an example to answer the question – Reinsbrain Apr 13 '21 at 16:35

How can I programmatically create a new cron job?

19 Answers19

Linked

Related