Re: Help with chpst -e

[Andras Korn <korn-supervise@elan.rulez.org>]

On Tue, Apr 15, 2008 at 12:05:02AM +0100, Robin Bowes wrote:

>>> # Get the service name
>>> logdir=`pwd`
>>> svcdir=${logdir%/log}
>>> SERVICE_NAME=${svcdir##*/}
>> Isn't this a bashism?
>
> Yes. I welcome suggestions how to do this in a more portable way.

How about

SERVICE_NAME=$(basename $(dirname $(readlink -f .)))

Of course, I don't know how portable readlink is.

>> Setting LOGUSER to the name of a directory is probably a
>> copy-paste mistake, I guess? :)
>
> No, it's based on the premise that each service is run by a user with the 
> same name. This is over-ridden by the contents of ./env/LOGUSER

Wow. I never considered using usernames with slashes in them, but of course,
why not?

>>> # Make sure the log dir exists
>>> if [ ! -e "${LOGDIR}" ]; then
>>>   mkdir -p "${LOGDIR}"
>>> fi
>> This fails if LOGDIR exists but is not a directory. (Of course, that
>> shouldn't happen.)
>
> True. I could use -d, but then do I test for $LOGDIR existing but not being 
> a directory? One has to stop somewhere...!

You're right, but at least "cd $LOGDIR || exit 1" fails in a predictable
way. :)

>>> # Set ownership & permissions on the log dir
>>> chown -R ${LOGUSER} "${LOGDIR}"
>> This gives ownership of the config file, if it exists, to svlogd. I don't
>> think that's good. 
>
> No, it gives ownership of the logdir to the user running the log process so 
> it can write to that dir.

chown -R is recursive, meaning that it will chown the svlogd 'config' file
which resides in the logdir to the user svlogd will run as.

I agree this only has theoretical significance.

Andras

-- 
                 Andras Korn <korn at chardonnay.math.bme.hu>
                 <http://chardonnay.math.bme.hu/~korn/>	QOTD:
              Smile... people will wonder what you've been up to.