[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: setting runlevel to single-user

Not to fan the fire or anything..

If daemons are terminating by themselves and it becomes necessary to restart them via a script or otherwise I would think that someone of reasonable intelligence (ie: able to tie their shoes) should be looking at the system in question.

Granted it'd be nice to be able to have a local data center janitor be able to restart a service if need be.. but maybe some services that are terminating are doing so for a reason that needs a more professional seeing-to.

And it begs asking at this point.. don't some people have little scripts to monitor (and restart?) the important things anyway?

Which springs to mind a funny story.. I knew a guy at an ISP who once wrote an external script that monitors a systems' ssh service and will ssh into the system to restart it (if sshd dies). It took some prodding (not cattle - unfortunately) for him to see the error of his ways.. :)

Greg Thomas wrote:


On Thu, Oct 28, 2004 at 07:10:24PM -0700, Greg Thomas wrote:

I realize this is just going to prolong this argument but as someone
like Brett Lymm,  has to tell clueless co-workers how to do things like
the above over a crappy cell connection it's much easier this way.
all I have to say.

Having dealt with this far too much myself, this argument doesn't hold any water. The only time you are doing this is when something is messed up, and 90% of the time that script is not going to work as a result.

In fact, depending on the rc scripts, it could very well hide the error
message that starting the daemon normally would give you, thus making it
harder to find the problem.

Its not hard to have someone grep rc.local, and then cut and paste a
line from there.  Certainly not any harder than having them run a
script, then run some more commands to see if the script actually
worked, then run some more to try to figure out why it didn't, then look
in the script to find out how its starting the daemon, and then run that

I probably should have described my environment.  I work in a very
MS-centric environment and I'm dealing with co-workers around the country
who have never logged into the Linux/FreeBSD/OpenBSD box that they
inherited.  They have no idea what grep or cat or how to cut and paste for
that matter.

If "/etc/rc.d/sshd start" doesn't work for them then they're in bigger
trouble then they realize.