Why does `timeout` not work with pipes?

Question

The following command line call of timeout (which makes no sense, just for testing reason) does not work as expected. It waits 10 seconds and does not stop the command from working after 3 seconds. Why ?

timeout 3 ls | sleep 10

@FrankieTheKneeMan I expected, `timeout` executes `ls | sleep 10`. Since `ls | sleep 10` will take at least 10 seconds, `timeout` should stop it. But `timeout` does not. — John Threepwood, Aug 14 '12 at 08:54

Shawn Chin · Accepted Answer · 2012-08-15T13:27:48.437

22

What your command is doing is running timeout 3 ls and piping its output to sleep 10. The sleep command is therefore not under the control of timeout and will always sleep for 10s.

Something like this would give the desired effect.

timeout 3 bash -c "ls | sleep 10"

edited Aug 15 '12 at 13:27

answered Aug 15 '12 at 12:06

Shawn Chin

84,080
19
162
191

1

Now I can see my mistake. Great workaround. Thank you very much (again), you really helped me a lot. – John Threepwood Aug 15 '12 at 13:25
Not a problem John. Glad I could help :) – Shawn Chin Aug 15 '12 at 13:28
Seems that is not enough when working with grep on a stream: `tail -f file | timeout 20s bash -c 'grep "a" | sort | uniq'` prints nothing.. – Ulysse BN Jun 17 '21 at 08:43

score 2 · Answer 2 · answered Aug 13 '12 at 22:20

2

The 'ls' command shouldn't be taking 3 seconds to run. What I think is happening is you are saying (1) timeout on ls after 3 seconds (again this isn't happening since ls shouldn't take anywhere near 3 seconds to run), then (2) pipe the results into sleep 10 which does not need further arguments than the number you are giving it. Thus ls happens, timeout doesn't matter, and bash sleeps for 10 seconds.

answered Aug 13 '12 at 22:20

mjgpy3

8,597
5
30
51

4

`ls` can take a very long time to run in a directory with a large number of files. – jordanm Aug 13 '12 at 22:36
1

Absolutely, you are correct. I made a few assumptions I shouldn't have in that answer, thank you for the insight. – mjgpy3 Aug 14 '12 at 02:22

score 2 · Answer 3 · answered Aug 13 '12 at 22:55

2

The only way I know how to get the effect you're after, is to put the piped commands into a separate file:

cat > script
ls | sleep 10
^D

timeout 3 sh script

answered Aug 13 '12 at 22:55

Thor

45,082
11
119
130

1

You can use process substitition to do that inline and avoid the temporary file -- `timeout 3 sh <(echo "ls | sleep 10")` – Shawn Chin Aug 15 '12 at 13:31

score 1 · Answer 4 · answered Mar 06 '20 at 19:28

1

It is enough to set the timeout on the last command of the pipeline:

# Exits after 3 seconds with code 124
ls | timeout 3 sleep 10

# Exits after 1 second with code 0
ls | timeout 3 sleep 1

answered Mar 06 '20 at 19:28

Kombajn zbożowy

8,755
3
28
60

"A duration of 0 disables the associated timeout. Note that the actual timeout duration is dependent on system conditions, which should be especially considered when specifying sub-second timeouts." Exit status: * 124 if COMMAND times out * 125 if ‘timeout’ itself fails * 126 if COMMAND is found but cannot be invoked * 127 if COMMAND cannot be found * 137 if COMMAND is sent the KILL(9) signal (128+9) * the exit status of COMMAND otherwise – ShpielMeister Aug 13 '21 at 04:09

Why does `timeout` not work with pipes?

4 Answers4

Linked