Re: startup and shutdown oracle in solaris 8

David Fitzjarrell wrote:

>
> I, also, have worked on a number of systems that only boot to run level
> 2, but it has been years since I have seen this. Possibly this is
> because I am usually working on large-scale systems with a user
> community spread across a large geographical area and there is a need
> for the system to attain run level 3. My advice applies to whichever
> run level is 'standard' for the UNIX system in question. Also, I don't
> usually find applications listed in /etc/rc?.d, just system services,
> although I am not stating in any fashion that this cannot occur. You
> are quite correct that any application that needs the database (and is
> started by initd) needs to be started after the instance and listener
> are properly started and closed prior to shutting down those same
> Oracle services.
>
> It matters not, really, which run level the system attains at boot
> time; what matters is that the symlinks are placed in the
> proper /etc/rc?.d directory to ensure that the Oracle processes are
> started when the system boots and stopped when the system is shut down.

Part of the problem that I have to deal with is that we often put implementations into customer environments where the customer has very little knowledge of how the system works. Because of this we tend to keep things simple and don't mess around with run levels very much. We simply boot up the machine and then it just sits there servicing the application and the only time that someone logs into the server itself is if something is being investigated or if the sys admin wants to perform routine maintenance chores.

This situation is also why we run the primary application packages from /etc/rc2.d - when the box boots it brings up the daemons, server processes, etc that server as the basis for the application. This way if worse comes to worse and the customer says that the application is not responding and we cannot log in remotely to investiage, we can just swallow real hard and tell the customer to turn the box off and on again and hope that chkroot and fsck don't come up with any errors, that the database comes up without any problems and that the application will be restored to normal function by the reboot. If all the work I did were in my own lab for internal purposes I'd probably screw with run levels more, but sometimes it is just impractical. Received on Fri Dec 01 2000 - 08:28:20 CST