Forum OpenACS Q&A: A tip for anyone installing Oracle 9iR2 on RHEL4....

1: A tip for anyone installing Oracle 9iR2 on RHEL4....

Posted by Janine Ohmer on 05/29/05 07:12 PM

I wanted to post this somewhere, and it doesn't seem really appropriate for the tip blog so here you go.

I did the abovementioned install last night, and at the configuration tools step the installer seemed to hang. I kept pressing the Next button and it would spin for a while and then stop. I figured it was stuck and left it, intending to figure it out today.

Well, this was running via X to one of our colocated servers, and overnight it chewed up over 4 mbits of bandwidth. Doing what I have no idea; I guess it must have continued to spin periodically. This is going to be one expensive Oracle install! :(

I didn't bother to try to figure out what was going on; I just exited the installer. I think it was finished anyway.

So, word to the wise - don't leave an Oracle installer unattended, it might try to run away with the family silver! :)

2: Re: A tip for anyone installing Oracle 9iR2 on RHEL4.... (response to 1)

Posted by Janine Ohmer on 06/01/05 09:15 PM

Here's an even bigger gotcha that people should be aware of.

There is apparently some very bad mojo that arises from a combination of the older glibc from the compat RPMs required by the Oracle install and the setting of LD_ASSUME_KERNEL to 2.4.19. Once these two things are in place, installing any RPM will cause the RPM database to be severely corrupted and unusable.

The problem is copiously described here:

https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=101603

But they seem to have thought they fixed it for RHEL4. Not so, as Mike can corrupt his RPM database at will just by setting LD_ASSUME_KERNEL.

This situation is relatively easy to avoid if you aren't lazy like me and don't set all your environment variables in /etc/profile. Only users who will be accessing Oracle need to have them set, and root generally isn't one of those users.