On 11.12.19 18:00, Alan Altmark wrote: > On Wednesday, 12/11/2019 at 04:18 GMT, "Hamilton, Robert" > <[email protected]> wrote: >> I'm interested in this too. I am following the read-only root > documentation, >> which is starting to show its age a little (RedPaper redp4322 is good, > but >> copyright 2008), but still has good stuff about gold instances, > maintenance and >> cloning. >> The other documents I have are the workbooks from some VM and Linux >> installation classes I've taken, but a lot of the preparation stuff is > skipped >> over. >> >> like, where and how to put an ISO. Nothing I've found in any cookbook > says how >> it was done, or what tools I can use to manage or unpack an ISO. BFS? > OMVS? RPI? >> >> I'm just interested in the latest doc. I'm trying to install ClefOS and > manage >> it as cleanly as possible. > > Mike and I are in agreement about read-only root: Don't do it. > Intellectually appealing, but impractical. It's good for a ysprog to play > around with such things to see what's *possible* with the available tech, > but just because a thing CAN be done doesn't mean it SHOULD be done.
Agreed. > > When we start to aggressively push process-changing Z tech into the Linux > admin world, it creates a rift. The x86 Linux admins wash their hands of > Z, leaving it to a sysprog who isn't part of the Linux admin team to do > Linux administration. And the sysprog is now left with another job and no > pay increase for doing it. You don't get invited to the Linux admin > parties and you're not part of strategy discussions. You get left behind > and sand gets kicked in your face. You get exasperated eyerolls and the > occasional "OK, Boomer." > > The real learning moment is when you simply ask your Linux admins, "How do > you install new Linux servers in your virtualized environments?" Based on > that answer, you figure out how Z can fit into that model. Just keep in > mind that better tech isn't always "best". By all means, ask them if > <some cool Z thing> would be of value to them, but be prepared to accept > "No" as an answer, no matter how much inherent mainframe coolness is > there. Right. It has to fit in the scheme of your company. I think a scheme that is used often in x86 is to use a cloud image and then use cloud init to provision the system. You basically use a daily image (e.g. https://cloud-images.ubuntu.com/bionic/current/ has daily builds of latest greatest 18.04 with all updates). You then use a config file passed through as virtual disk to provision the system. While this works natively with kvm, there are also documents that describe cloud-init for z/VM. (I have not tested that with z/VM so I will just assume that this will work as well) A kind of similar thing is used in coreos called ignition, so a base image that is customized with config seems to be common in the container/k8s world as well. Christian ---------------------------------------------------------------------- For LINUX-390 subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO LINUX-390 or visit http://www2.marist.edu/htbin/wlvindex?LINUX-390
