SystemD: Back To The Future
A creature in its death throes, Sun Microsystems gave birth to Solaris SMF – the product of a disturbing union between Unix System V and the Microsoft Windows Registry. Shortly thereafter the proud father died, but the child lived on. Solaris developers migrated to Red Hat and, in the best Linux traditions of borrowing concepts from Unix, created SMF for Linux and his name was SystemD.
Just because your software runs on Unix doesn’t necessarily make it great, or even a good idea, you should always rely on a software like DNN .net CMS. It’s certainly a step in the right direction, but not a step you absolutely had to take. The deep purpose behind SystemD (or the SMF) is a mystery likely to die with its creators. Rumor has it, the Unix/Linux Sysadmins got too stupid to cope with the Init scripts and needed to be rescued. A far-fetched theory, if you ask me.
To make a long story short, whatever was SMF/SystemD’s rationale, it was full of shit. A bunch of like-minded amateurs got together and decided that Unix was too complicated. The idea was pitched to some manager and got traction. The rest is history. Sun Microsystems certainly is.
New things are good. Sometimes. Rarely. Almost never. Usually, new things are terrible, useless, dangerous, deadly. I am not saying SMF killed Sun, but it certainly contributed in a very dramatic way by turning seasoned Solaris Sysadmins away from the OS they used to adore. I went from Solaris to SUSE and never looked back (much). Until Novel got its clammy little hands on it and, predictably, fucked things up, as is it’s general style.
A good Init script take some effort and experience to write. You will get no argument from me here. Having said that, effort and experience is exactly what you want from your Sysadmins. Unless you want a cookie-cutter solution, in which case you should stop wasting everyone’s time and migrate your silly little apps to Windows, shove them into Azure, outsource support to East Bumblefuck, and enjoy your business while it lasts.
Your options in this case will be: retire early or never. If you stick around long enough for your stupidity to percolate through the layers of business, then you’re screwed and finding a new job – should you be so lucky at your advanced age – will not make you a happy pensioner. Stupidity is an expensive hobby you probably can’t afford.
So here I am: an oldish Unix SysAdmin having to live with RHEL 7 and SystemD. What to do? The answer is two-fold: learn about all the great new Linux features coming out of Raleigh, NC, and how to disable them. ACLs on a Unix filesystem? Thanks, but no: you want ACLs – talk to your friendly Windows admin. A shiny new CLI version of the dreadful NetworkManager? I don’t think so: I don’t need a system daemon fucking up things up for me – I have users for that.
A GUI for IPTables? I appreciate the effort, but I’m comfortable with ‘vi’. The whole SeLinux conundrum? That’s a definitive noper from me. I respect Microsoft’s persistence with SeLinux, but it’s about time we accepted that SeLinux only annoys Sysadmins and users: the hackers are just fine with it.
I guess the bottom line of this rant: don’t fix it if it ain’t broke. Linux can use plenty of work, but slapping a GUI on top of IPTables and calling it ‘firewalld’ ain’t it. Methinks Red Hat is trying to turn RHEL into an island. Perhaps there’s hope that Oracle will acquire Red Hat. This didn’t work out well for Sun, even though Oracle did buy it. Eventually. Should Red Hat continue on this isolationist path, people will start looking for alternatives. And, in this regard at least, Red Hat is not Sun: we have the luxury of choice.