Andrei POPESCU wrote:
On Mi, 05 nov 14, 09:32:57, Miles Fidelman wrote:3. What other options/initiatives are you aware of that you've discarded or otherwise are not considering, and why?Not sure why you're not considering Debian GNU/kFreeBSD. Due to the BSD kernel it's guaranteed to not have systemd, not even the libraries, and it's still Debian.
That's a good point. That was an oversight on my part. I have been looking at it (and continuing to do so), but also keeping my eye on some of the concerns about its possible demise.
- I'm specifically concerned about the impacts of upgrading to Jessie, in terms of having to re-wire collections of packages, configuration, initialization scripts, etc. that are supporting several production systems. All reports I've seen suggest that I'll end up with servers out of commission for a while, and more than a few sleepless nights.In my opinion it's a pity you rely only on reports from others and don't try it out for yourself.
Call it due diligence.When someone (or lots of people) are yelling "fire" - I generally don't run into the building to check it out for myself. I start looking for a fire extinguisher and calling the fire department.
It's one thing if this was about evaluating and selecting bleeding edge technology (say a new cloud stack), but this is an operating system that we're using in production. What counts is operational stability and minimum impact when making changes. Call it operational conservatism - I NEVER deploy or upgrade anything until its been through several releases and the number of bug reports die down. THEN I review specs, release notes, howtos, and so forth; plan a transition; and start testing on a staging server.
Beyond that, our production servers have become somewhat highly tailored and configured over time. The systemd ecosystem represents some serious re-wiring of core "plumbing" to an extent that I can't see any way that it won't break some of our configurations in subtle ways (every major upgrade does this, it's to be expected, but systemd looks like it will do this in spades). Finding those impacts involves a lot more than simply installing a copy of jessie on bare iron and seeing if it boots. It involves replicating our operational environment (high-availability cluster), and stress testing. And even then, no amount of testing is going to identify all the nits that will result in sleepless nights tracking down the cause of a midnight crash.
It's a simple matter of time allocation and prudence. Before "buying" anything - software, cars, whatever - I look at specs, documentation, reports, reviews, tests, and so forth. If they raise alarm bells, I'm not going to go out for a test drive or start building up servers. There are enough other things to do in a day.
Right now, everything I've been hearing and reading is telling me to run from jessie as fast as I can - and that I should be investing time and effort in testing (more like evaluating) alternative platforms.
-- In theory, there is no difference between theory and practice. In practice, there is. .... Yogi Berra