Shell script wait for background command

[rogeriovinhal]

Hello, I am writing a script, but there is something I need that I can't find a way to do it...

I need to make a command in background "command1 &" and then somewhere in the script I need to wait for it to finish before I do command2. Basically, I need this:



start script
...
command1 &
...
more commands to run parallel to command1
...
wait for command1 to finish
command2
end of the script


I've looked a lot of pages, but I couldn't make this simple task to be done. I've seen something about a 'wait' command with the pid process number, but that didn't work also (or I couldn't make it work).

Can anybody help me?

EDIT

Nevermind, I solved it.

To those that may be interested, the problem was that the command1 was a wine call, but it seems that wine calls returns a exit signal even before it stops working, so "wait" didn't wait for it to finish properly. I solved this by executing the 'wine blablabla' command inside a subshell that should return a value at background "$( wine blabla ) &", so now wait really worked.

[tonybaqain]

1. when you run a command in the BG using the & notation, it is now in the jobs daemon. so jobs -l will display the jobs with information about them, as follows:


jobs -l
[2]- 9647 Running ping 192.168.11.27 >&status &
[3]+ 9656 Exit 1 df -hs &>/dev/null


2. the "wait" command usage is as follows
wait [n]
where n is the PID to wait for

this example clearifies the usage:


.
.
.
[command] &
wait_command_pid=`jobs -l | awk '{print $2}'`
.
.
.
wait $wait_command_pid
.
.
.


jobs -l | awk '{print $2} will print out the pid number only, which is the second delimitted word at the output of the job itself.

[tonybaqain]

futhur more, to be more specific, and to let you use more than one bg command
do the following :

instead of jobs -l | awk '{print $2}' use jobs -l | grep criteria | awk '{print $2}', but you have to be aware that not to use the same command twice, or even similer commands that cann't be grepped.

[tkharris]

You could do that, or you could do this:


tribeca:~ harris$ bash foo.sh ; cat foo.sh
Pid = 372
#!/bin/bash

echo "Foobar" >/dev/null & EPID=$!
wait $EPID
echo "Pid = $EPID"
tribeca:~ harris$


=D
$! contains the PID of the last called process. :)

[komputes]

wait for command1 to finish
You can also hard-code how long you want the script to wait before going onto the next process by doing

command1
sleep 5
command2

In this example script, command1 will run, wait 5 seconds, and run command2

[ghostdog74]

instead of jobs -l | awk '{print $2}' use jobs -l | grep criteria | awk '{print $2}', but you have to be aware that not to use the same command twice, or even similer commands that cann't be grepped.

just FYI, grep is redundant

jobs -l | awk '/criteria/{print $2}'