I’m documenting how to perform a custom MySQL install from scratch and I aiming for an “offline” installation using yum. So the intention is to download the rpm files for the community sever and then call yum to install the server, without actually having to configure the any repository repository or having to connect to it.
So I’m basically following the documentation and downloading:
That appears to be correct, though there might be some other dependencies which are commonly available in the main OS repo. I would use rpm -Uvh to install the packages locally, which would prevent any external fetch. You can also install the ‘download-only’ tool for yum, then tell it to install mysql. It will download the RPMs but not install. Then you can take a look at all the packages it needs.
Indeed. So yum prompts for some dependencies, but since the host is already connected to the Red Hat repository, it would download them from RedHat. Would that create any conflict in that yum is pulling rpms needed to resolve those dependencies from RedHat?
The reason why we want to do it offline is to be able to pick whichever version we have and place it in our depot, from where we would then deploy to other on-prem servers. But having redhat resolve missing dependencies can be advantageous. Alternatively we will have to download all those RPMs instead, which may be different on every version.
I like the ‘download-only’ tool for yum that you mentioned, as we could deploy that on out satellite server/depo and then have yum download without install. Then we push them to the DB server and do an rpm -Uvh. But for that, I will have to check with the Unix team to make sure this does not meddle around with the default yum configuration they already have with Redhat