On Thu, Sep 14, 2017 at 11:39 PM, Robert Fisher <[email protected]> wrote:
> > Yes. That 'distributed SMF' thing, too ;) > > > > For me that would be an ability to see and influence service states on > remote envs (including neighboring other zones), so e.g. if a tracked > instance of database got up somewhere, > > > start this appserver here (or the opposite for orderly shutdowns). I had > some clumsy PoCs back in the day, but nothing good to share except a few > ideas ;( > > I wrote a thing years ago that let you query and manage service states > over HTTP. It's not very sophisticated: it ends up shelling out to the > `svc*` commands, the security model is embarrassingly weak, but it did > work. It's not been touched in years, and shows its age, but it's still > around. https://github.com/snltd/SexyMF. It probably won't run unmodified > on a modern node platform. > Yes, I remember that now. Are there other tools in that vein still around? > The one thing I liked that you-know-who added to Solaris 11, was RAD, > which takes the concept much further, giving you an API to (pretty much) > the whole OS. I'd love to see something like that in SmartOS, used > primarily to make one SMF know what the others are doing, and react > accordingly. Fire off a message to other zones/hosts when the state of a > service changes; have those SMFs listen for and act on that message. You > could do a lot with that. > There was an early version of rad shipped with OpenSolaris, but it didn't do much. In the wider sense, we've lost a lot of the administrative tools. We don't have rad, visual panels has gone, the various predecessors (admintool, adminsuite, symon, etc) are no more. For IPS, the gui went. There's clearly an opportunity to improve matters dramatically in this area, because it's so vacant at the moment. > On the broader discussion, innovation is happening further up the stack > now, because most real-world work is being done several layers of > abstraction up from the OS. Change happens when people are deeply unhappy > with something, and I think people are pretty happy with their OS now. > That's assuming they ever even think about their OS. Which, generally, > they don't. > Cause or effect? I think you're right in terms of what's actually happening, but I suspect people are working further up the stack because that's where they can, not because they should. I find the idea that the OS is good enough to be hugely troubling. > This, I think, is what's sealing the fate of non-linux operating systems. > It's not necessarily that people *prefer* linux, or could begin to argue > why it might be in some sense *better*, they just don't care what's at the > bottom of the stack. > > The OS is becoming as unloved and out-of-sight-out-of-mind as the BIOS. > Where's the innovation in the BIOS? Your BIOS sucks, but how often do you > think about it? Where's the real-world pressure to innovate and disrupt? > *illumos-discuss* | Archives > <https://illumos.topicbox.com/groups/discuss/discussions/T83f198c8597cf8e3-M9872c2a387b5ba2383e88fc7> > | Powered by Topicbox <https://topicbox.com> -- -Peter Tribble http://www.petertribble.co.uk/ - http://ptribble.blogspot.com/ ------------------------------------------ illumos-discuss Archives: https://illumos.topicbox.com/groups/discuss/discussions/T83f198c8597cf8e3-M1abbedaee7ace64cc86d1453 Powered by Topicbox: https://topicbox.com
