Sleep process problem

[u856100]

Hi All,

I have a script that is respawned by the inittab. When the script is already running, sometimes we need to stop it for some work to be done. Unfortunately not all of us have the root password, so instead of waking the UNIX administrator up at 0'clocks in the morning, we put the script to sleep by editing the header (adding sleep 6000). When the script is killed and subsequently respawned, the script obviously spawns a sleep process. The problem we are having, is that when the sleep process 'only' is killed, and the script starts up again, eventually it will die (after a few hours), and respawn 'again'. We do not know what is causing this, and are obviously keen to find out.

many thanks
John

[Mark Greene_1]

After you modify the script, you have to kill HUP init to get it to reread the script file, because it *may* still have the old copy of the script in memory.

mark

[Nicolas Dumeige]

Editing the script header will only be taken into account for the next run. So apparently, you count on the process to die on its own (that's why you add the sleep 6000), but you are suprised when it die after a few hours. So, what's bother you is that it isn't spawn again after it dies ?

[John Carr_2]

Hi

Mark I dont think John can kill HUP if he is not root !

John you could ask your sys admin to install sudo and give you the ability to kill HUP

:-) John.

[Elmar P. Kolkman]

John, have you considered a sleep 3600 will sleep for 1 hour, so a sleep 6000 will only sleep for less then 2 hours?!

[u856100]

Hi Chaps, thanks for the responses, they are appreciated!!

The sleep values etc are irrelevant in my case as I am just preventing the script from running so I can prevent users from logging onto a weblogic instance (started, stopped by the script). The weird thing is that when I create the sleep process, kill it so that the script starts running from where it left off, the application crashes and restarts again after a time period. This time period is much larger (2-3 times) than the sleep value I specified, so it can't be that.

This has happened a couple of times now.

thanks again chaps

[Colin Topliss]

Are you able to get this script changed?

If the script is basically looping, I'd do something along the lines of getting the script to check for the presence of a file. If it finds a file, sleep. If it doesn't, then continue. The flow would be:

init program loop
if file exists
sleep 500
else
process normally
fi
endloop

Just touch the file its looking for to start the sleep process, then delete it when you want processing to continue.

Col.

[Mark Grant · Accepted solution]

I don't think modifying a running script will always have the results you hope for. The idea sounds clever but there must be a better way.

Can't the script be made loop arouns a "sleep" itself if, for example a file exists and you can touch and remove that file.

I do have another alternative for you which may be a bit of overkill for what you want.

THe attached script behaves like init in that it can "respawn" processes. However, with this script, processes are given named run levels. A process runs (and is respawned) within one or more run levels. Thus you could respawn your script and when you want to stop it, you just change to a run level that the script isn't in.

As an encore, the script also does various alerting things and monitors itself.

[Mark Grant]

Pressed the wrong button!

Here's the attachment

[Mark Grant]

You'll also need an example config file

[Paula J Frazer-Campbell]

John

Lateral thinking:

Get the sys admin to give you full permision on the script.

if script is called fred then create an empty one called fred-do-nothing.

Routine will them be

cp fred fred.bak
cp fred-do-nothing fred

When finished reverse it

cp fred.back fred

Reason
Init will not report errors if the script exists.


Simple

Paula