Re: A clean way to introduce delay between scripts in /etc/init.d?

[Stephen Samuel <samnospam@bcgreen.com>]

One thing you can do is check every few seconds to see if the script has 
completed.
You could, for example, have bind9 remove a blocking file when it is 
starting, and recreate it when it is done (or vice versa).  Have your 
later script check for the existence of the file when it starts.

It's also a good thing to put a timeout into the loop, so that you don't 
block permanently.

I have a script snippet that does something similar. It checks to see if 
pump has assigned an IP address in the wanted subnet, and then uses that 
address if it is assigned (or times out after 8 attempts).

In this case, you can use the value of $emsg to decide what to do after 
this loop (I just print it).
The question for success here is whether the read gets enough words to 
fill 'num' with the
last octet of the IP address from 'ifconfig -a'

--
# the network address:
NETIP=192.168.211
dmsg="waiting for dhcp"
emsg="-n ''"

for i in 1 2 3 4 5 6 7 8   splat  ; do
       read a b c d num remaining  <<-EOF
       ` ifconfig -a | grep inet | grep -F $NETIP | tr . ' ' `
       EOF
       if [ -n "$num" ]  ; then
               [[ -z "$emsg" ]] && emsg="-n" || emsg=" done! "
               netpart=$b.$c.$d
               subnet=$d
               break
       else
               echo -n "$dmsg "
               dmsg=" ... "
               emsg=" timeout!"
               [ "splat" = "$i" ] || sleep 5
       fi
done
echo  $emsg "<$num>"

exit

========================



LEE, Yui-wah (Clement) wrote:
> Hi,
>
> What is a clean way to introduce a delay between the
> scripts in /etc/init.d ?
>
> I saw a problem that /etc/rc2.d/S27bind9 started before
> /etc/rc2.d/S25ifplugd actually completed all the tasks
> ...
>
> Specifically, ifplugd, with the help of resolvconf,
> would generate the file /var/run/bind/named.options,
> which would be used by bind9.  Unfortunately by the
> time bind9 ran, the file was not yet generated.
>
> I could force a delay between the two scripts by
> introducing another script (run at S26) that does
> nothing but sleep for a fixed time (30 seconds).  That
> seemed to work around the problem.  But I wonder if
> there is a cleaner solution.
>
> Ideally, the delay should not be fixed but dynamic.
> That is, a later script can have some means to know for
> sure that an earlier script has completed all the tasks
> already.
>   
--
Stephen Samuel +1(778)861-7641             samnospam@bcgreen.com
		   http://www.bcgreen.com/
  Powerful committed communication. Transformation touching
    the jewel within each person and bringing it to light.